Content uploaded by Sofie Beier
Author content
All content in this area was uploaded by Sofie Beier on Jun 16, 2022
Content may be subject to copyright.
Sofie Beier
designing for legibility
Reading Letters
designing for legibility
Copyright © 2011 Sofie Beier and BIS Publishers
Sofie Beier
Ovink
Publishers
Het Sieraad
Postjesweg 1
1057 DT Amsterdam
The Netherlands
T (+) 31 (0)20 515 02 30
F (+) 31 (0)20 515 02 39
bis@bispublishers.nl
www.bispublishers.nl
ISBN 978-90-6369-271-1
All rights reserved.
No part of this publication may be reproduced or transmitted in any
form or by any means, electronic or mechanical, including photocopy,
recording or any information storage and retrieval system, without
permission in writing from the copyright owners.
Reading Letters was partly funded by the Danish Centre for Design Research.
Text and design
Typeface
Publisher Reading Letters
Sofie Beier
Publishers
designing for legibility
Acknowledgment
Parts of the book are based on my PhD thesis ‘Typeface Legibility: towards defining
familiarity’, written while affiliated with the Royal College of Art, London, UK.
During the development of the book, a number of people have kindly informed
me on various aspects of their own expertise. In this regard I would especially
like to thank Dan Reynolds for his helpful feedback on the final draft and for
providing me with images of blackletter type. I would also like to thank Kevin
Larson and Mary Dyson for general advice on scientific legibility research. I am
very grateful for the input given by James Mosley enlightening me on historical
matters, and for the help provided by Jan Middendorp pointing me towards use-
ful typographical images.
Furthermore, I would like to thank Chris Burke for putting me in contact with the
Tschichold estate, Gerard Unger for providing drawings of Dwiggins’ puppet, and
Eric Kindel for the test material applied by The Graphic Information Research Unit.
Last but not least, a big thanks to all the designers and foundries who so kindly
have supplied me with images of their work. Without these, the book would
never have been the same.
Sofie Beier
Introduction ····················
1. Test methods ··················
Readers’ preferences
Continuous reading
The search task
Visual accuracy threshold
2. Understanding reading ···········
Letter identification
Word superiority effect
Word wholes
Parts, wholes and context
3. Legibility in histoty ·············
4. Theories on letter structure ········
Edward Johnston
Gerrit Noordzij
Frank E. Blokland
5. Stroke and contrast in history ·····
The Old Style stroke
Romain du Roi
The Baskerville stroke
The Didone stroke
The geometric stroke
6. The individual letter ·············
Internal letter relations
Scientific results
Designers’ ideas
7. Type for text sizes ··············
Block text
Proportions
The M-formula
Ink and printing
Type for screen
Scaling
8. Type for distance viewing ········
Compensation for loss of details
Proportions
9. The capitals ·················
Roman inscriptions
Relative width
Which is more legible?
10. Sans or serif? ················
Arguments in favour of the serif
Scientific findings
Familiarity of the sans serif
11. Hanging and ranging numerals ·····
12. The italic evolution ···········
The Cancelleresca italic
The development of the Didone italic
Cursive italic & sloped roman
13. Letter spacing ···············
14. Familiarity in history ···········
Reading and writing in the Middle Ages
Blackletter and Latin types
The Civilité type
John Baskerville and his peers
Didone and the French Revolution
15. The validation of legibility ········
Index ························
Contents
Introduction
This book will, on the one hand, help type designers create highly legible type-
faces and, on the other hand, help graphic designers determine the optimal
typeface for a given project.
Few of us will appreciate if the typeface we read is legible; however, we
quickly notice if it is not. Creating type for optimal legibility is therefore a
thankless task, since readers only register your failures. For instance, typefaces
presented under difficult reading conditions, such as small point sizes in low-
quality newspaper print or street and building signs viewed from afar, need to
be created in specific ways to function at their very best.
To understand the topic in depth, two very different areas of expertise
have been consulted. One area is that of punchcutters and designers who,
through their experience, possess useful knowledge that can help us better
understand the various aspects of the matter; the other is that of academic
reading research, where a significant number of relevant scientific studies
have been carried out over the years.
Type designers have a subtle understanding for details and nuances that
seem difficult to test in a laboratory, but many of the theories and ideas
presented by designers do in fact lend themselves to laboratory testing.
Some theories have already been verified, while others have been rejected.
The outcome of this research has yet to be made widely available to design-
ers. Consequently, many designers make assumptions without really knowing
whether they are right or wrong. In a synergy between the two topics, this
book will evaluate typeface legibility from different angles in an effort
to provide useful information that can, hopefully, be transferred to practice.
7
arm
ascender
aperture
x-height
cap height
descender
loop
tail
contrast stem
axis
counter
crossbar
ear
shoulder
teardrop
terminal
junction
spine eye
Constituent parts of letters
1
Test methods
Scientific studies on legibility-related matters have been carried out in a
number of ways over the years. Many of the methods have been criticised
for being insufficient. An argument often raised is that we must understand
legibility fully before attempting to study it. One implication of this would be
that all existing test methods are ineffective, since reading is such a complex
process that no single method can ever produce sufficiently useful results.
A common criticism is that a reader who is placed in a laboratory setting
will always be aware of the action of reading, which means that obtaining
a realistic measurement becomes problematic. The claim is that the human
mind is too complicated for any valuable information to be extracted during
a laboratory test. Whenever one aspect is tested, a range of other factors will
inevitably influence the subject, and the experiment is bound to lead to inad-
equate findings. However, studies carried out in a natural environment allow
for too many uncontrollable variables, while a laboratory setting, on the other
hand, makes it easier to control and isolate the many correlating factors that
play a role in everyday life.
A central point of criticism that is voiced in the legibility debate is that
different test methods produce different results. Legibility researcher Miles A.
Tinker1 compared typefaces in terms of visibility under reduced illumina-
tion, perceptibility at a distance, speed of reading, and the reader’s opinion
about the most legible type. In this study, he found little agreement between
the results of the four test methods. In the measurement of reduced illumi-
nation, bolder types performed better than lighter ones, a finding that had
much in common with the test of perceptibility at a distance. Tinker also
found that the reader’s opinion was less compatible with speed of reading
than the two other aspects, and that readers in general judged types that
perform well in distance studies to be best for comfortable reading. Instead
of viewing this as a setback for the prospects of scientific legibility investi-
gation, one might view the different findings as an indication of something
more useful. Comparisons such as these reveal that typeface legibility is not
a universal issue, where one feature or set of features improves legibility in all
reading conditions. In other words, the level of legibility for a given typeface is
not constant but varies, depending on the situation in which it is observed.
Figure 1.1. Different test methods
produce different results. The findings
by Miles A. Tinker when he exposed the
same typefaces to different test methods
showed a large difference in performance1.
The typefaces are illustrated with digital
fonts that are similar in style to the metal
types applied by Tinker.
Legibility ranking
( = best performance)
Visibility un-
der reduced
illumination
Perceptibility
at a distance
Speed of
reading
Readers’
opinion
of legibility
-
-
(illustrated with Bookman Old Style)
(illustrated with Century Old Style)
(illustrated with Adobe Caslon Pro)
Continuous reading
Designers often argue that book typefaces should only be tested in running
text, as this is, after all, how the type is going to be read. The issue is not,
however, quite as straightforward as it may seem. Comparing two columns of
text set in different typefaces raises a range of potential dilemmas. Leading
and spacing in the text always interact with one another, an issue that is
particularly evident when matching two designs of different x-height (Fig. .).
If the leading is kept constant, one of the two is likely to be at an advantage.
If, on the other hand, the leading is adjusted to give the typefaces a visual
similarity, one text might take up more space on the page than the other,
which may introduce a bias.
A common goal in studies of continuous reading is to measure reading
speed. It may, however, be problematic to assume automatically that fast
reading equals high legibility. Perhaps speed should not be the goal. Maybe
when we read a highly legible text, the type will make us read with less
effort rather than increasing the speed of reading. This notion is supported
by a study2 that compared text with different margins; the study found that
although the speed of reading was reduced when readers were exposed to the
text with larger margin, they had a better understanding of the content under
this condition.
Another central criticism in relation to tests of reading speed is the fre-
quent lack of significance in the measured time differences between the fonts
tested. Unfortunately, this does not necessarily mean that there is no differ-
ence; more likely, the test method simply is not sensitive enough to detect
any variation.
To test continuous reading, there are several methods to choose from. One
method is to look at oral reading. Here, the participant reads a text aloud, and
the researcher records the number of errors or the time it took. A problem
with this approach is that the situation is unnatural for most adults. It can
be difficult to determine whether mistakes are based on errors of identifica-
tion or errors of interpretation or memory. When reading aloud, we often use
similar words or restructure the text to improve the flow; this does not mean
that we actually decode the text that way. Another consideration is that oral
reading leads to a higher frequency of fixations on the line of text, and that
oral reading speed is about half that of silent reading3. Furthermore, because
the eye is faster than the voice when we read aloud, only substantial perfor-
mance differences will show up in the test results.
Figure 1.2. What leading can do to a
text. Leading can pose a challenge when
comparing typefaces of different x-height.
The optimal leading for a typeface with a
small x-height is not always the same as the
optimal leading value for one with a large
x-height. Illustrated with the typefaces Bran-
don Grotesque by Hannes von Döhren, and DS
Musee by Dino dos Santos, both set in 22-point
size, with a line height of 22. Note the differ-
ence in the white areas.
Brandon Grotesque, / DS Musee, /
The search task
Another possible approach is to measure readers’ ability to locate specific
words or letters in a text. Such methods were applied in studies carried out
by the research unit working at the Royal College of Art6 in the s and s.
Their reason for choosing this method was a dislike for testing the readers’
comprehension, as they judged the comprehension method as being influ-
enced by too many unaccountable variables that were not related to the
visual properties of the text.
The problem with the search task method, however, is that it is more
closely related to scanning than to actual reading. In addition, since most
reading situations do not allow us to know the exact words we are about to
read, the method is not a very close approximation of real-life situations.
When participants know what to look for, the interrelation between the
process of reading the word and the processes of reading the individual letter
is dominated more by the word process than is the case in more natural read-
ing situations (see more in Chapter ). That being said, the search task method
is very useful for studying participants’ ability to identify a target object
located among other elements.
Based on the problems related to measuring oral reading, it might make sense
to study silent reading instead. The challenge here is how to measure read-
ing when others cannot hear it. One way around this challenge is to test for
comprehension and check whether readers have understood the text that
they read. This does, however, lead to a new range of issues that need to
be addressed. How, for instance, can we ensure that participants all have
the same degree of interest in the topic of the text? If they find the topic
boring, will that influence their concentration, and will their comprehension
suffer as a result? Will we in fact be testing the participants’ intelligence or
simply their experience of being in a test situation? Furthermore, it has been
demonstrated that high-frequency words such as ‘the’ are read faster than
other three-letter words4, and that sentences in the active voice are recog-
nised faster than sentences in the passive voice5. If the experiment involves
comparing two different texts set in two different fonts, the level of high-fre-
quency words and the structure of the individual texts are likely to influence
the outcome of the study. Then again, if all participants are tested on all type
conditions, and if the test material is counterbalanced between conditions,
these issues should not cause problems.
Figure 1.3. Eye tracking. The measure-
ment of eye movements during continuous
reading can provide researchers with useful
information on lengths and durations of
saccades and fixations. Illustrated with the
typeface Stella, designed by Mário Feliciano.
Visual accuracy threshold
In this approach, the focus is on letter and word identification, while com-
prehension is not a priority. Participants in tests based purely on perception
tend to be unaffected by the awkwardness of the test situation. As we know,
optometrists can make rather accurate vision measurements in laboratory set-
tings. That is because word recognition occurs on an automatic level and is
therefore unaffected by the surroundings. This is confirmed by the famous phe-
nomenon discovered in by J. Ridley Stroop7. Asking participants to name
the print colour when presented with words such as ‘Yellow’ in green print and
‘Blue’ in red print, Stroop demonstrated that most participants found this to
be very difficult; thus, the study concluded that we recognise words even when
it would help us to ignore them (Fig. .).
There are several methods for measuring visual accuracy; one way is to briefly
expose the participant to the stimulus. After this rapid exposure, which is so
short that the eye is unable to move from one fixation to another, the partici-
pant is asked to identify the presented material. Because of the single fixa-
tion, an obvious risk is that the test method might vary too much from a more
regular reading situation. On the other hand, it can be argued that the differ-
ence is not that significant, since the eyes are relatively stable when fixating
Figure 1.5. The Stroop effect. The task is to
name the print colour of the word ‘Blue’ when
it is printed in red, or the ink colour of the
word ‘Yellow’ when printed in blue, etc.
Illustrated with the typeface Ovink by Sofie Beier.
BLUE
YELLOW
RED
GREEN
Figure 1.4. Search task test material. Test
material applied by The Graphic Information
Research Unit working at The Royal College of
Art in the s and s. Participants were
asked to locate the words on the left in the
text shown on the right.
In another accuracy-based test method, the focus is on the visual angle
between type size and its distance to the eye. It is a highly relevant method
for studying the perception of signage in general. The tests carried out in
connection with the development of the latest typefaces for American road
signs10 are good examples of this (Fig. . and .). One may choose to meas-
ure the minimum distance at which a participant can read the stimuli, or one
may calculate the number of stimuli the participant recognises from a given
distance.
It has often been discussed whether this test method has any relevance
in relation to continuous reading. Applying a distance method in their experi-
ments, optometrist James E. Sheedy and colleagues11 argued that if a type-
face is identifiable at a small point size, the larger the difference in size
between the lowest identifiable size and the size applied for reading, the
higher the legibility - yet the researchers also recognise that although it
seems logical that this feature would improve reading performance, that has
yet to be established.
While many issues can be raised with regard to the validation of this test
method when applied to the study of distance perception accuracy of a sig-
nage typeface, the method has rarely been questioned.
Figure 1.7. Distance study. A picture showing
the test situation in a legibility study carried
out by the UK Road Research Laboratory in .
The test was carried out following a heated
public debate about the typeface created for
the British road signs by Jock Kinneir and his
assistant Margaret Calvert12.
on a stimulus, both in short exposures and in normal reading situations. It
thus seems reasonable to assume that the perceptual processes would be the
same. Yet this may not always be the case. It appears that humans sometimes
process the same information differently, depending on the task they set out
to perform9. In the short-exposure method, the participant mentally prepares
for the task of perceiving the material in one fixation. In normal reading,
the fixation is a part of a larger process and is not treated as a specific task.
However, the method is useful for studying the legibility of individual charac-
ters or words and for identifying specific features for further investigation.
Figure 1.6. Testing short exposure with a
mask8. To control the time frame in which
the image will appear, it is helpful to let the
stimulus be followed by some form of mask
consisting of random dots or lines. That will
remove the afterimage from the retina. a
a
1. Tinker, M.A. (1944) ‘Criteria for Determining the
Readability of Type Faces’, Journal of Educational Psychol-
ogy, vol.35, pp.385-396.
2. Chaparro, B.S., Baker, J.R., Shaikh, A.D., Hull, S.
& Brady, L. (2004) Reading Online Text: A Comparison
of Four White Space Layouts [online], Usability News,
vol.6(2), available from:
www.surl.org/usabilitynews/archives.asp.
3. Rayner, K. & Pollatsek, A. (1989) The Psychology of
Reading, Englewood Cliffs, NJ:Prentice-Hall International.
4. O’Regan, J.K. (1979) ‘Moment to Moment Control
of Eye Saccades as a Function of Textual Parameters in
Reading’, in A. Kolers, M.E. Wrolstad & H. Bouma, (eds),
Processing of Visible Language, New York: Plenum Press,
vol.1, pp.49-60.
5. Forster, K.I., & Olbrei, I. (1973) ‘Semantic heuristics
and syntactic analysis’, Cognition: International Journal of
Cognitive Psychology, vol.2(3), pp.319–347.
6. Reynolds, L. (2007) ‘The Graphic Information
Research Unit: a pioneer of typographic research’, Typo-
graphy Papers, vol.7, pp.115-137.
7. Stroop, J.R. (1935) ‘Studies of interference in se-
rial verbal reactions’, Journal of Experimental Psychology,
vol.18, pp.643-661.
8. Beier, S. & Larson, K. (2010) ‘Design Improvements
for Frequently Misrecognized Letters’, Information Design
Journal, vol.18(2), pp.118-137.
9. McConkie, G.W. (1983) ‘Eye Movements and
Perception during Reading’, in: K. Rayner, (ed), Eye Move-
ments in Reading: Perceptual and Language Processes, New
York: Academic Press, pp.65-96.
10. Garvey, M., Pietrucha, M.T. & Meeker, D. (1997)
‘Effects of Font and Capitalization on Legibility of Guide
Signs’, Transportation Research Record, National Academy
Press, Washington D.C., no.1605, pp.73-79.
Hawkins, H.G., Picha, D.L., Wooldridge, M.D., Green,
F.K. & Brinkmeyer, G.R. (1999) ‘Performance comparison
of three freeway guide sign alphabets’, Transportation
Research Record, no.1692, pp.9-16.
11. Sheedy, J.E., Subbaram, M.V., Zimmerman, A.B. &
Hayes, J.R. (2005) ‘Legibility and the Letter Superiority
Effect’, Human Factors: The Journal of the Human Factors
and Ergonomics Society, vol.47(4), pp.797-815.
12. Lund, O. (2003) ‘The public debate on Jock Kin-
neir’s road sign alphabet’, Typography Papers, 5, 103-126
13. Shaikh, D. (2009) ‘Know Your Typefaces! Semantic
Differential Presentation of 40 Onscreen Typefaces’,
Usability News, vol.11(2), available online from:
www.surl.org/usabilitynews/archives.asp.
Femnne Mascune
Soft Hard
Delicate Rugged
Relaxed Stff
Beautfu Ugy
Expensive Cheap
Good Bad
Passve Actve
Cam Excted
Sow Fast
Quet Loud
Sad Happy
Weak Strong
Coo Warm
Od Young
egbe Legbe
Mean
-
Femnne Mascune
Soft Hard
Delicate Rugged
Relaxed Stff
Beautfu Ugy
Expensive Cheap
Good Bad
Passve Actve
Cam Excted
Sow Fast
Quet Loud
Sad Happy
Weak Strong
Coo Warm
Od Young
egbe Legbe
Mean
-
Readers’ preferences
To understand the reader’s experience of a given reading situation, scientists
ask participants to rank their preferences. It is a useful method for determin-
ing how a certain group of people responds to a certain typeface, or whether
the typeface appeals to the public in general. It is, however, of little use if
one is interested in the legibility of the typeface or in any other performance-
related aspect. If participants are asked to pick the most legible typeface, in
all likelihood their choice will reflect their own ideas about aesthetics, not
what is actually the most legible. What people think, and what people do, are
two very different things. Another key aspect when asking for opinions in a
test situation is that most of us would like to please the experimenter and
avoid the risk of hurting his or her feelings. While it is difficult to improve
one’s own performance in terms of reading speed or accuracy in identifying
letters from afar, it is quite easy to answer preference questions favourably
if one can guess the experimenter’s preference. The main purpose of this
method therefore lies in the elaboration of the reader’s opinion and motiva-
tions, while it is of limited value when performance is of interest.
Figure 1.8. Readers’ opinions about typeface
personality. The researchers asked partici-
pants to grade a range of typefaces according
to personalities.
References
2
Understanding
Reading
Letter identification
There are two main theories on the process of letter identification. The first
theory is the template matching theory. The essential idea here is that for
each letter of the alphabet, the brain has stored a basic template of the
letterform. As we perceive a new shape, the brain flips through a series of
templates to find the best match. This is also the theory espoused by type
designer Adrian Frutiger3, who compares the function of reading to a keyhole
and its key, where the reader locates the basic skeleton form of the letter,
which then fits like a key into the keyhole and triggers identification.
However, the main problem with this theory is explaining how we are
capable of recognising the wide variations in typefaces and handwriting that
we actually manage to handle. If this were indeed how the system works, the
Figure 2.2. Feature comparison. The fea-
ture comparison theory states that when
recognising a letter, we first recognise the
individual features of the letters, then
combine the features to eliminate related
letters, and eventually end up with a final
identification. Illustrated with the typeface
Facit by Tim Ahrens.
To define what actually happens when we perceive words and letters is not an
altogether easy task. Over the years, scientists have come up with a number
of theories to explain the act of reading. These theories range from the one
extreme stating that we perceive the words exclusively as wholes without
recognising the individual letters1 to the other extreme stating that reading is
based solely on a letter-by-letter recognition process2.
Although we have yet to fully understand how the brain works when read-
ing, we do have some idea of the general process.
A
no
no
no
yes
no
no
yes
yes
Figure 2.1. Template matching. According
to the template matching theory, all readers
have some form of basic master of each letter
stored in memory. The question is how we
manage to identify letters in very different
typefaces and handwriting styles. Illustrated
with Vogue Paris, Ornamenta, A2 BrewTypeDisplay,
and A2 Monday, all by Henrik Kubel.
surrounding templates should be of no importance. However, if we analyse
the results by applying the feature comparison theory, searching for the letter
‘’ in a visually related character group means searching among a range of
similar features, which would offer a plausible reason for the slowing down of
the search process.
However, feature comparison might not be the sole explanation. Another
study6 revealed that fixating an image to the retina, so that when the eye
moves the image moves along with the eye, eventually causes the image to
disappear (Fig. .). The study found that complex stimuli sometimes disap-
peared and reappeared as a whole and sometimes vanished in fragments.
This suggests that both the template matching and the feature comparison
processes play a role, and that they are interrelated in the workings of our
perceptual system.
Figure 2.4. Disappearing images. When we
try to hold our eyes still, a slight tremor will
always occur; this is essential for our vision.
Research shows6 that if an image is fixated
to the eye so that it follows the movement
of this tremor, the image will fade and
disappear from the retina. This effect may
apply to either parts or whole images.
brain would have to have a separate template for a flamboyant ‘ ’ and a
simple sans serif ‘A’, and for all variations of handwriting. Even if the brain
has some form of clean-up process for letter shapes, it nevertheless seems
doubtful that a system like this would be able to decide which parts of a
character shape are essential and which are not. The shortcomings of the first
theory led to the feature comparison theory.
Instead of perceiving the whole character, this next theory states that
the brain decodes the individual features of the character one by one. This
analytic process is based on a perception of the characters as a range of
disparate features that are gradually combined, until a stage of identification
has been reached.
One argument in favour of the feature comparison theory comes from
a study on the visual system of cats4. By projecting different forms of pat-
terns onto different regions of the cat’s retina, the researchers demonstrated
that the cortical cells in the cat’s visual system fired differently according
to whether the stimulus being processed was a horizontal line, a vertical line
or a curve. Obviously, there are differences between feline and human visual
perception; however, it is commonly accepted that this identification process
of lines and curves in the visual cortex is rather similar between the two spe-
cies. Another finding that supports the feature comparison theory comes from
Ulric Neisser5, who showed that in a search task, test subjects found it easier
to locate the letter ‘’ in a group of visually unrelated characters ()
than in a group of visually related characters () (Fig. .). If we read
in the way suggested by the template matching theory, the results of this
study should show no difference between the unrelated and related character
groups. If the eye is searching for one particular template, the shapes of the
ODUGQR
QCDUGO
CQOGRD
QUGCDR
URDGQO
GRUQDO
DUZGRO
UCGROD
DQRCGU
QDOCGU
CGUROQ
OCDURQ
UOCGQD
RGQCOU
GRUDQO
OCURDO
DUCOQG
CGRDQU
UDRCOQ
GQCORU
GOQUCD
URDCGO
GODRQC
IVMXEW
EWVMIX
EXWMVI
IXEMWV
VXWEMI
MXVEWI
XVWMEI
MWXVIE
VIMEXW
EXVWIM
VWMIEX
VMWIEX
XVWMEI
XMEWIV
MXIVEW
VEWMIX
IVWMEX
IEVMWX
WVZMXE
XEMIWV
WXIMEV
EMWIVX
IVEMXW
Figure 2.3. Find the ‘z’. The study by Ulric
Neisser found that it is easier to locate the
letter ‘z’ in the left column of letters with
visually different features than in the right
column of letters with similar features.
Word wholes
Many internet users have encountered a circulating text referring to a
research project which found that “it deosn’t mttaer waht oredr the ltteers in a
wrod are, the olny iprmoetnt tihng is taht the frist and lsat ltteres are at the rghit
pclae”. The text further concludes that “Tihs is bcuseae we do not raed ervey
lteter by it slef but the wrod as a wlohe”. Testing this jumbled word effect, sci-
entists10 have found that reading speed generally slows down when letters are
transposed. If we take a closer look at the text in question, most of the swap-
ping occurs between neighbouring letters, none of the swapped letters create
new words, and all the function words such as it, the, in, a, are, and is stay the
same, which makes it easier to guess the content of the sentence. Moreover,
if reading is based on word wholes alone – as the text claims – the shifting
of ascending and descending letters would break up the word shape and thus
undermine our ability to identify the word. One can further assume that the
phenomenon will be less prominent when applied to languages with many
compound words, such as the Scandinavian languages, German or Dutch.
If we do read by word wholes, words set in mIxEd cAsEs should slow down
our reading rate dramatically. Scientists11 have found, both in reading aloud
and in a search task study, that words set in mixed cases where the char-
acters retained their original x- or cap-height sizes did not perform well.
Figure 2.6. Mixed case. Words set in upper-
case and lowercase letters where the height
of the letters is adjusted to the same level
(top) are easier to recognise than words set
in regularly sized mixed letters (middle).
Illustrated with the typeface Neue Swift by
Gerard Unger.
amueon
hamburgefont
HamBuRGeFonT
Word superiority effect
In , James McKeen Cattell showed that after a short exposure at a close
reading distance, participants were more likely to identify single words than
single letters. The phenomenon is known as the word superiority effect. Later,
in , another scientist7 recreated the experiment with a few adjustments,
changing the experiment to a forced choice between two target letters that
were presented after the stimulus. This was done in such a way that both
the two alternative choices would make up a word in connection with parts
of the stimulus word (Fig. .). The study found that a target letter in a real
word was more accurately recognised than either single letters or a target
letter in a nonsense word8. On the basis of these findings, it can be hypoth-
esised that reading is based on the long-term recollection of words and word
patterns. However, other studies show that this is not exactly the case. When
pronounceable nonsense words such as ‘mave’, or ‘reet’ are included in experi-
ments based on the forced choice method, studies9 have found that these
words are – in most cases – recognised far better than unpronounceable non-
sense words such as ‘ftgy’, or ‘ojhl’. This indicates that the word superiority
effect is a result of letter combinations rather than familiar word patterns.
Figure 2.5. Forced-choice method. A word,
non-word or letter is briefly presented to
the participants at a normal reading dis-
tance. Next, the participants are presented
with a choice between two letters and
asked to identify which of the letters was
part of the stimulus. Illustrated with the
typeface Beckett by Henrik Kubel.
Parts, wholes and context
Most of the studies reviewed so far do not contradict the basic ideas of the
Parallel Letter Recognition () model13 (Fig. .). The model contains three
basic levels, the first being the feature detector level. As described earlier, the
process at this stage involves recognising the features of the individual let-
ters, such as horizontal, vertical, curved and diagonal lines. This information
is then passed on to the letter detector level. If an ‘o’ is part of the stimu-
lus material, the letter detectors for ‘o’ would be active in combination with
letter detectors for other related shapes such as ‘c’ and ‘e’. The task for the
letter detectors is to locate the letter with the highest number of common
features to match the information received from the feature detector level.
The final level involves the word detectors, which in a similar process identify
the component features (letters) and combine them into words. What hap-
pens next on the word detector level is not yet fully understood14. However, it
appears that a second process takes place on the word detector level, consist-
ing of the top-down input of some sort of lexical stimulus based on context,
word wholes and word parts. This operation in turn proceeds further down to
the letter level in a parallel process.
This parallel top-down and bottom-up processing in the model
explains the word superiority effect. While single letters have to be identi-
fied exclusively by information received from the letter detectors, words are
decoded on the basis of information from both letter detectors and word
detectors, and therefore, presumably, words will have a higher recognition
rate than individual letters. When the perceived word is not identified in the
word lexicon, we have to spell out the word, relying on the letter level. If a few
letters cannot be identified, the collaboration between the word lexicon and
Figure 2.8. Reading without recognising all
letters. In the corporate identity created by
e-types and GD for the Danish Designers, the
letter ‘d’ is replaced by a square. In spite of
knocking out one of the important letters of
the Danish language, the text still appears
readable.
Word detectors
Letter detectors
Feature detectors
Stimulus detectors
However, it was also found that words set in mixed case where the height of
uppercase and lowercase letters was adjusted to the same level performed the
same as words set only in uppercase or lowercase letters (Fig. .). Maybe the
problem with the unadjusted mixed-case words has more to do with familiar-
ity than with word wholes. Another research project12 looked into this and
found that the performance of mixed-case nonsense words and mixed-case
regular words was equally impaired. If in fact our reading is based entirely on
whole-word recognition, the performance difference of familiar regular words
changed to mixed-case would be radically different from the performance differ-
ence of nonsense words changed to mixed-case, since the nonsense words do
not contain any familiar word shapes, neither in normal case nor in mixed case.
Figure 2.7. Parallel Letter Recognition
(PLR) model. According to this model, we
read by the parallel operation of a bottom-
up process, where we identify letter fea-
tures, and a bottom-down process, where we
identify words. Illustrated with the typeface
Stella by Mário Feliciano.
the letter lexicon will serve to identify the word. The model further explains
the jumbled word effect; it seems reasonable to assume that as long as there
is no phonetic confusion, the collaboration between the predictability deliv-
ered by the top-down process and the detailed information of the bottom-up
process will be capable of identifying swapped characters, as long as they are
not placed too far apart.
So what is the internal relationship between these different processes that
influence reading? A recent study by Pelli and Tilman looked into the matter
by isolating the three mental processes of letter-by-letter, word-wholes, and
sentence-context recognition. The scientists measured reading rates in oral
and silent reading of printed text and on text presented in a rapid serial visual
process, where the text is displayed word by word at the same position on a
screen. The manipulations shown in Figure . were tested both one at a time
and in combinations. The study found that the letter function is the strong-
est factor, accounting for about % of the adult reading rate; the sentence
function came in second %; while the word function was the weakest of
the three, accounting for only % of the reading rate. The three processes
appear to operate in collaboration.
Combining these findings with the ideas of the model gives us a good
indication of the different kinds of operations that take place in the reading
process. The collective research suggests that the functions of letter, word,
and context detectors support each other by approaching the reading mat-
ter from different angles. Although highly dependent on the other detectors,
the function of identifying the individual letter comes across as the strongest
single factor.
Figure 2.9. The test material of the Pelli
and Tillman study. Illustrating the three
kinds of ‘knock-outs’, finding that readers
were most troubled when reading the ‘letter
knock-out’.
Knock-out Examples
Sentence Contribute others. the of Reading measured
Word ThIs text AlTeRnAtEs iN CaSe.
Letter Tbis sartcrec bes Ictfan suhsfitufas.
1. Huey, E.B. (1908) The Psychology and Pedagogy of
Reading, New York: The Macmillan Company.
2. Sperling, G. (1963) ‘A model for visual memory
tasks’, Human Factors, vol.5, pp.19-31.
Gough, P.B. (1972) ‘One second of reading’, in: J.F.
Kavanagh & I.G. Mattingly, (eds.), Language by ear and by
eye. Cambridge, MA: The MIT Press, pp.331-358.
3. Hunziker, H.J. (1998) ‘Typographic work of Adrian
Frutiger’, Serif: the magazine of type & typography, vol.6,
pp.32-43.
4. Hubel, D.H. & Wiesel, T.N. (1962) ‘Receptive fields,
binocular interaction and functional architecture in
the cat’s visual cortex’, Journal of Physiology, vol.160,
pp.106-154.
5. Neisser, U. (1967) Cognitive Psychology, New York:
Appleton-Century-Crofts.
6. Pritchard, R.M. (1961) ‘Stabilized images on the
retina’, Scientific American, vol.204(6), pp.72-78.
7. Reicher, G.M. (1969) ‘Perceptual recognition as a
function of meaningfulness of stimulus material’, Jour-
nal of Experimental Psychology, vol.81(2), pp.275-280.
8. Wheeler, D.D. (1970) ‘Processes in Word Recogni-
tion’, Cognitive Psychology, vol.1, pp.59-85.
9. Aderman, D. & Smith, E.E. (1971) ‘Expectancy as a
Determinant of Functional Units in Perceptual Recogni-
tion’, Cognitive Psychology, vol.2, pp.117-129.
Baron J. & Thurston, I. (1973) ‘An Analysis of the
Word-Superiority Effect’, Cognitive Psychology, 4, 207-228.
McClelland, J.L. & Johnston, J.C. (1977) ‘The role of
familiar units in perception of words and nonwords’,
Perception & Psychophysics, vol.22(3), pp.249-261.
Carr, T.H., Davidson, B.J. & Hawkins, H.L. (1978) ‘Per-
ceptual Flexibility in Word Recognition: Strategies Affect
Orthographic Computation But Not Lexical Access’,
Journal of Experimental Psychology: Human Perception and
Performance, vol.4(4), pp.674-690.
10. Rayner, K., White, S., Johnson, R. & Liversedge, S.
(2006) ‘Raeding Wrods With Jubmled Lettres; There Is a
Cost’, Psychological Science, vol.17(3), pp.192-193.
11. Smith, F., Lott, D. & Cronnell, B. (1969) ‘The
effect of type size and case alternation on word
identification’, American journal of Psychology, vol.82(2),
pp.248-253.
12. Adams, M.J. (1979) ‘Models of Word Recognition’,
Cognitive Psychology, vol.11, pp.133-176.
13. McClelland, J.L. & Johnston, J.C. (1977) ‘The role
of familiar units in perception of words and nonwords’,
Perception & Psychophysics, vol.22(3), pp.249-261.
Rayner, K. & Pollatsek, A. (1989) The Psychology of
Reading, Englewood Cliffs, NJ:Prentice-Hall International
Larson, K. (2004) The Science of Word Recognition
[online], Microsoft Typography, available from: www.
microsoft.com/typography/ctfonts/WordRecognition.aspx
(an abridged version in: Eye, no.52, pp.74-77)
14. McClelland, J.L. & Rumelhart, D.E. (1981) ‘An
Interactive Activation Model of Context Effects in Letter
Perception: Part 1. An Account of Basic Findings’, Psycho-
logical Review, vol.88(5), pp.375-407.
Rumelhart, D.E. & McClelland, J.L. (1982) ‘An Interac-
tive Activation Model of Context Effects in Letter
Perception: Part 2. The Contextual Enhancement Effect
and Some Tests and Extensions of the Model’, Psychol-
ogy Review, vol.89(1), pp.60-94.
Paap, K.R. & Noel, R.W. (1991) ‘Dual-route models of
print and sound: still a good horse race’, Psychological
Research, vol.53, pp.13-24.
15. Pelli, D.G. & Tillman, K.A. (2007) Parts, Wholes, and
Context in reading: A Triple Dissociation [online], PLoS ONE
2(8): e680, New York University, Available from:
psych.nyu.edu/pelli.
References
3
Legibility
in history
it function as an ‘l’. At the time of Smith’s writing, the Caslon Foundry was
the most dominant in Britain; however, the serifs on the ascenders of ‘d’
and ‘b’ in these types are quite different from the serifs on the descenders
of ‘q’ and ‘p’; furthermore the ‘u’, unlike the ‘n’, has no serifs on the right
site of the stems (Fig. .). Based on this, Smith’s ideas were most likely not
something William Caslon himself would have approved of. Probably for the
same reason, the edition of The Printer’s Grammar no longer included the
relevant paragraph.
One of the most influential typefounders of his time, Pierre Simon Fournier
(-), applied a commonly used practice of reusing the same counter-
punch when creating the inner form of letter groups like ‘b’, ‘d’, ‘p’, ‘q’ and
‘h’, ‘n’, ‘u’1. This method makes it a bit of a challenge to differentiate the
individual characters. Fournier therefore had to find another solution, which
he described in the preliminary note to his Modèles des Caractères of :
“I also have given the corners rather a squarer cut, and this I have done to some
of the lower-case as well, and removed a certain roundness which was observable
at the junction of vertical and horizontal strokes; this gives them an appearance
of greater independence, separates the one from the other, and makes them more
evidently distinct”2. By focusing on the overall features, Fournier managed to
create letters of high individuality without undermining letter regularity.
The Didot family of printers and typefounders eventually became the lead-
ing figures in the French typography movement. As seen in the works of their
contemporary typefounder and printer Giambattista Bodoni (-) in
Italy, the Didot types evolved into designs with a high degree of internal simi-
larity between letters. Although later criticised for their poor legibility, they
were generally accepted by the public (see more in Chapter ). One contem-
porary critic, however, was the Frenchman Sobre, who in explained his
antipathy as follows: “The truth is that Garamond was careful to emphasise those
parts of the shape of his types which distinguish them from one another—the ties
for instance—while Didot emphasised those parts of the shapes of his types which
are common to all”. Sobre further theorised that in the ‘u’ and ‘n’ designed by
Claude Garamond (d. ), the focus is on the top section of the ‘n’ and the
Figure 3.2. Counterpunch. There are two
ways of creating the counters of letters in
punchcutting. One is by digging the counter
out with a graver (top); the other is by strik-
ing a counterpunch into the punch (bottom).
Figure 3.3. A large-size Fournier typeface.
Note the difference between the aggressive
diagonal shoulders of the characters ‘h’, ‘m’,
and ‘u’, compared with the round, softer
shapes of ‘e’, ’o’ and ‘e’.
Except for a number of printing houses famous for their exquisite work and
craftsmanship, the main focus for most printers prior to the th century
was on building a profitable business. The legibility of the type they applied
was less essential. In the edition of The Printer’s Grammar, the author
John Smith described how printers living too far away from typefounders for a
regular resupply could replace certain letters with others when the proper let-
ters became broken. Listing these letters, Smith explains how the characters
b-q, p-d and n-u may be rotated to replace one another; ‘e’ could be changed
to ‘c’ by cutting out the central stroke; cutting the ascender off an ‘h’ could
turn the character into an ‘n’, and cutting the ‘n’ part off the ‘h’ could make
Figure 3.1. A lack of respect for character
shapes. If printers are short on certain sorts,
author John Smith suggested in 1755 that they
simply rotate or cut off elements of other let-
ters. Illustrated with the revival Williams Caslon
Text, created by William Berkson.
32 33
Based on the assumption that legibility is defined by the smallest amount
of correlation in the surface of the characters, Lucien Alphonse Legros and
John Cameron Grant5, one an engineer and the other a writer, set out in
to test the legibility of a range of typefaces (Fig. .). The work was carried
out by calculating the number of units (each one thousandth of an inch) in
a square covering each character in a typeface. The theory was that high leg-
ibility would be equal to a low number of shared units across letters within
the same design.
It is apparent that the repetition of shapes and low contrast, which is
dominant in the applied sans serif, performs rather badly under the circum-
stances described above, whereas the more organic Blackfriars offers a much
Figure 3.5. Early type by Bodoni. Show-
ing the Transitional style features of ‘c’,
‘e’ and ‘o’. Furthermore, the teardrop of ‘f’
is much heavier than any other teardrop,
giving emphasis to the ascending parts that
differentiate ‘f’ from ‘t’. Image from Bodoni’s
Anacreontis Teii Odaria of 1789.
bottom section of the ‘u’, and therefore “you cannot for a moment be in doubt
as to which it is”; he further argued that in the ‘u’ and ‘n’ of Didot’s types, the
connecting parts have such thin hairlines that “you have to use discernment to
avoid confusing them”3.
In , five years after the death of Giambattista Bodoni, Bodoni’s widow
finished and published his Manuale Tipografico, which showed specimens of his
many typefaces. Here Bodoni discussed the regularity and harmony within the
units of the letters and stated that “the standardisation of every thing which is
not in itself distinctive, and the accentuation, so far as is possible, of the neces-
sary marks of differentiation, will impart to all the letters a certain schematic
regularity”4. Bodoni’s later work did not demonstrate a high concern for letter
differentiation, yet a look at his earlier typefaces reveals features that appear
to some extent to bear out the statement above. In the example from
shown in Figure ., the difference in axis between the vertical ‘o’ and the
diagonal ‘c’ and ‘e’ emphasises a distinction between these characters. Also,
the top of the ‘f’ is rather heavy, which helps differentiate the letter from ‘t’.
Figure 3.4. Lack of differentiation in the
Didot fonts. A comparison between the
Garamond ‘n’ and ‘u’ (left), and the Didot ‘n’
and ‘u’ (right) demonstrates the argument of
Citizen Sobre that the characters of the Didot
fonts add emphasis to similar features while
downplaying features that are different. This
is also evident in the letters ‘c’ and ‘e’, whose
distinctiveness in the Didone faces relies
solely on the hairline crossbar of the ‘e’. Here
demonstrated with Adobe’s Garamond Premier Pro
and H&FJ Didot.
34 35
better result. The authors recommended the kind of serif applied in Old Style
typefaces, as they found that “a heavy serif adds considerably to the non-
coincident areas of the il, un, and bh pairs of lowercase characters”6. This study is
interesting in that it has a clear mathematical approach to the argument of
enhancing the individuality of characters. However, it fails to elaborate on the
central issue of cohesiveness within the characters of a font. Later research7
has demonstrated that reading performance worsens when the letters come
from a mix of two fonts of different typefaces, compared to when they come
from a single font. A designer who followed the Legros & Grant approach
blindly would obtain high scores in terms of creating a typeface with charac-
ters that have no features in common; that does not, however, mean that the
typeface would perform well in a normal reading situation, where some level
of uniformity must be expected.
Figure 3.6. Legros & Grant: sans serif and
Blackfriar typefaces. By superimposing
the letters on one another, Legros & Grant
defined high legibility as the smallest cor-
relating area between letters.
1. Carter, H. (1930) Fournier on Typefounding: the Text
of the Manuel Typographique (1764-1766), Translated into
English, London: Soncino Press.
2. Carter, H. (1930) Fournier on Typefounding: the Text
of the Manuel Typographique (1764-1766), Translated into
English, London: Soncino Press, pp.289-290.
3. Updike, D.B. (1928) ‘Address by the Citizen Sobre
on the types of Gillé, and the Discourse of Berlier before
the Council of the Five Hundred in the Year 7 of the
Revolution’, The Fleuron, vol.6, pp.167-184, p.181.
4. Haddon Craftsmen (1937) A Primer of Types: Giam-
battista Bodoni 1740-1813, Camden, p.3.
5. Legros, L.A. & Grant, J.C. (1916) Typographical
printing surfaces: The technology and mechanism of their
production, London: Longmans, Green, and Co.
6. Legros, L.A. & Grant, J.C. (1916) Typographical print-
ing surfaces: The technology and mechanism of their produc-
tion, London: Longmans, Green, and Co., p.164.
7. Sanocki, T. (1988) ‘Font Regularity Constraints
on the Process of Letter Recognition’, Journal of Experi-
mental Psychology: Human Perception and Performance,
vol.14(3), pp.472-480.
References
36
4
Theories on
letter
structure
Figure 4.2. Johnston’s Foundation hand.
One of the writing hands recommended by
Johnston is based on an old English manu-
script written with a broad-nib pen held
at a -degree angle to the page.
Figure 4.1. The Johnston Sans. Edward
Johnston was primarily a calligrapher. He
was never really interested in the reproduc-
tion aspect of typography, a factor that
influenced the sans serif typeface he cre-
ated for the London Underground in .
Johnston believed that the pen should
dictate the nature of the stroke; he was not
accustomed to the need for optical adjust-
ments of curves and junctions, an issue that
is of keen interest to most type designers.
He did not consult with an experienced ty-
pographer on the job and stayed firmly with
the concept of creating a ‘block letter’ type
with a stroke width that was completely
uniform throughout.
Edward Johnston1
Edward Johnston’s (-) foremost ambition was to uncover the
many forgotten techniques of the early manuscripts. In his book Writing &
Illuminating, & Lettering of , Johnston took the reader through the various
calligraphic styles of his discoveries and guided the student on handling the
pen in the writing hands that he had investigated.
Although he often stressed the importance of following the specified
structural line in each written piece, Johnston also recognised that on certain
occasions the penman would find it necessary to break with the predeter-
mined style and adjust the pen to create a particular letterform that would
not be possible otherwise. He did, however, prefer types characterised by
simplicity and tool-made forms, and he emphasised the importance of hold-
ing the pen at a constant angle. He argued that when properly executed,
the shapes of the pen would determine the shapes of the letters. Johnston
did not generally recommend the technique of rotating the pen, and he did
not consider the pressure of the pen an essential influence on stroke width.
Following the norms of the Arts & Craft movement, Johnston centred his
attention on the early manuscript books of the broad-nib pen, which were
all written before copperplate engraving and the pointed-nib pen had had an
impact on formal penmanship.
Figure 4.4. The double pencil. To study the
structure of the broad-nib, Edward Johnston
recommended tying two pencils firmly together
with the points at the same level. In use,
the two points reproduce the outlines of the
broad-nib stroke. Without the ink to conceal
the inner structure, the double pencil can reveal
the internal anatomy of a letter, thus clearly
demonstrating what happens inside the curves
of the pen stroke.
Figure 4.5. Weight in calligraphy. Follow-
ing the nature of calligraphy, variations in
weight within a letter will be dictated by
variations in stroke thickness and contrast.
Altering the weight when writing with a pen
not only alters the relationship between the
black and white areas of the letter, it also
alters the basic shapes, as heavier weights
have a more abrupt shift between wide and
narrow sections.
Note the difference in the inner shapes of
the ‘o’ that occurs when the weight is al-
tered by changing the width of the pen nib.
Figure 4.3. The angle of the pen.
Working with the broad-nib pen, Johnston
demonstrated how holding the pen at dif-
ferent angles can result in very different
letter shapes.
Angle: The angle of the pen to the writing
surface defines the location of the thick
and thin parts of the strokes and thus also
the stress of the letters. A vertical stress
is created with a pen nib held at a hori-
zontal angle to the paper. This produces
letters with thin horizontal strokes and
thick vertical strokes. A diagonal stress is
created with a slanted pen, which results
in types where the thin parts of the round
shapes are located in the upper left area
and lower right area.
Weight: The weight of a calligraphic hand
is determined by the size of the letters
and the width of the pen nib. Any change
in weight invariably influences the shape
of the letters as well: the heavier the
weight, the more the pen will dominate
the letter shapes.
Skeleton: The skeleton is a letter stripped
of all its unessential parts, yet with its
basic structure, characteristics, and pro-
portions of the essential form intact.
Edward Johnston’s three primary writing conditions.
Gerrit Noordzij2
Like Edward Johnston, typographer and teacher Gerrit Noordzij (b. ) bases
his theories entirely on personal studies of historical manuscripts. Advocating
the importance of researchers making their own first-hand discoveries, he
rarely refers to conventional terms and concepts from type history. Noordzij
criticises authors of the ‘official history of typography’ for copying each
other’s statements and failing to investigate the topics of their writings
personally.
As in Edward Johnston’s work, the core of Noordzij’s theory is that all
typographical letter shapes originate in the writing hand, and that the shape
of the tool and the movement of the writing define the letterforms. According
to Noordzij, a stroke has two dimensions, one being the width of the pen nib,
and the other being the direction in which the pen is drawn. Noordzij’s basic
argument is that type should always be created with an overlay of a written
stroke as a guideline for the designer. He further emphasises the importance
of focusing on the black and white of the letter shapes simultaneously and
argues against the practice of creating type based on outline alone.
The debate about the merits of Noordzij’s theories has been ongoing since
his ideas were first published in the s. Critics express concern about
Noordzij’s disregard for the influence of printing methods on typographical
development, pointing out that ever since the advent of printing in the late
th century, book typography has not been created directly with a pen.
The limitation of Noordzij’s approach becomes apparent in regard to the
wide range of constructed geometric typefaces. Although Noordzij sees con-
struction as a form of writing – where shapes are created from a sequence of
strokes rather than being created from single strokes – geometrically based
typefaces do not fit naturally into this framework. Nevertheless, Noordzij’s
theories have proven their merits in the many well-crafted designs produced
by his students over the years.
Figure 4.6. Three typeface conditions.
According to Noordzij, all typefaces and let-
ter styles can be created out of an inter-
action between the three forms of stress
– Translation, Expansion and Rotation. The
categories are defined as follows:
Translation: The stroke is made with a
broad-nib pen held at a steady angle to the
page. The contrast of the strokes changes
when the direction of the pen changes, the
maximum width of the stroke is equal to the
width of the pen, and the minimum width is
equal to the sides of the nib. The stress of
the letter is diagonal.
Expansion: The stroke is made with a
pointed-nib pen held at a steady vertical an-
gle to the page. The contrast of the strokes
changes when the level of pressure on the
pen changes, the maximum width of the
stroke is defined by maximum pressure to
the pen, and the minimum width is defined
by minimum pressure to the pen. The stress
of the letter is vertical.
Rotation: The contrast of the stroke
changes when the pen is rotated. The width
of the stroke is defined as the counterpoint;
the angle of the counterpoint varies when
the writer rotates the pen. Illustrations by
Noordzij.
Figure 4.7. The Noordzij cube. Noordzij sees his ideas as an
alternative to the conventional typeface classification systems,
all of which are rooted in historical references. Based on his
concept of translation and expansion mixed with decreases or
increases in contrast, Noordzij has created a cube that shows
the gradual transformation between different elements influ-
encing the letter. The point here is that an Old Style typeface
and a Humanistic sans serif typeface should not be categorised
as very far apart but rather be viewed as closely connected,
as they differ only in terms of contrast and the presence or
absence of serifs.
Figure 4.8. Gerrit Noordzij on the pen and the writing surface.
Another of Noordzij’s ideas has to do with a distinction between
types according to the pen’s relationship with the writing surface.
This distinction falls into the following two categories:
Running construction: The writer keeps the pen on the writing
surface, using both upstrokes and downstrokes, as seen in cursive
and script faces.
Interrupted construction: The writer lifts the pen from the
writing surface, using only the downstroke of the pen, as seen in
roman faces.
Combining these categories with the influences of translation and
expansion produces a matrix with four units.
Illustration by Noordzij.
Translation Expansion
Frank E. Blokland3
As one of Gerrit Noordzij’s former students, the Dutchman Frank E. Blokland
(b. ) was initially influenced by Noordzij’s approach to letter structure.
Over the years, however, Blokland has developed his own theories on the matter.
Blokland argues an influence of an underlying structure of patterns and
regularisation on the Renaissance type forms. To prove this idea, he studied
a number of punches and printed materials found at the Museum Plantin-
Moretus, including works by the Frenchman Claude Garamond (d. ) and
the Fleming Van den Keere (-). What he found was that the original
sixteenth-century types have a clear standardisation of width, which enables
a grouping of lowercase letters as shown in Figure .. Speculating that the
typographical unit called the ‘em-square’ originally followed the width of the
lowercase ‘m’, Blokland investigated whether the width of this letter initially
had an influence on the vertical proportions of the type. His findings show
that it is in fact highly possible (see the Proportional System in Figure .).
Blokland further demonstrates that like the lowercase alphabet, the uppercase
letters appear to have been fitted to the dimension of the lowercase ‘m’ (Fig. .).
Figure 4.11. The sum of letter parts.
In his description of the issues that influ-
ence letterforms, Frank E. Blokland classifies
the individual factors into the following
categories.
. Harmonic system describes the morpholog-
ical relationship between letterforms as the
result of, for instance, the movements by
the pen, its rotation, and pressure. However,
the morphological relationship does not by
definition have to find its origin in writ-
ing, an example of this is the geometrically
constructed letterforms.
. Proportional system describes the vertical
and horizontal proportions of characters,
including the relationship between the x-
height and the ascenders/descenders, and
between uppercase and lowercase letters.
. Relational system describes the boldness
of the characters as a relative parameter in
relation to the height and contrast of the
character. The category is similar to Edward
Johnston’s description of ‘Weight in cal-
ligraphy’ mentioned in Figure ., with an
additional focus on the effect provided by
the thickness of the nib.
. Rhythmic system defines the intervals of
stems and counters and thus also the spac-
ing of the characters.
. Formalisation describes the decreasing
level of influence the pen stroke has on the
letterform.
. Idiom describes the influence the style
of the type designer has on the character’s
shapes.
Figure 4.10. The proportions of the
lowercase ‘m’ as a standard for uppercase
width. The uppercase letters of the early
Renaissance were likely designed to fit the
width of the lowercase ‘m’. Illustrated with
the revival Adobe Jenson by Robert Slimbach.
Figure 4.9. Uniform width of lowercase
letter groups. Blokland shows that in the
th century, the following letter groups
had identical width: b-d-g-h-n-u-o-p-q-
v-y-fi, i-j-l, a-c-e, and r-s-t. Illustrated by
Blokland with Garamond’s/Van den Kere’s
‘Moyen Cannon Romain’.
.
.
.
.
.
.
Type design
We know that some of the early punchcutters started their careers in other
crafts working as engravers and goldsmiths. Due to their background, Blokland
finds it plausible that they were aware of conventions like the golden section,
which was widely applied in the Renaissance world of arts, and thus they must
have applied these ideas in their own work as well. As he superimposes the
golden section on both Gutenberg’s textura type and on several revivals and
original Renaissance typefaces, Blokland shows a strong concurrence between
the golden section and the various type forms (Fig. .).
The theories presented by Blokland partially contradict the beliefs of many
type historians who argue that, due to the small scale of the punches, the
early punchcutters used the judgment of their eyes as the driving force in
the production of type. With his findings, Blokland points out that geometric
measurements and standardisation must have had a strong influence on the
production of type as well, and that this essential knowledge has been lost
in history somewhere along the line. That being said, he also recognises that
some of the standardisations he proposes might have been too complex for
the early punchcutters to apply. Yet, Blokland’s studies seem to confirm the
notion that the early punchcutters were using more than their eyes in the
construction of letter proportions, and that they somehow systematised their
work by applying structures of mathematical calculations to their type.
References
Figure 4.12. The golden section.
Blokland demonstrates that the golden
section appears to have had an influence on
the proportions of letters produced by the
early punchcutters. Illustrated with the revival
Adobe Jenson by Robert Slimbach.
1. For more on Johnston’s theories see:
Johnston, E. (1906) Writing & Illuminating, & Lettering,
London: John Hogg.
Johnston, E. (1938) Manuscript & Inscription Letters,
London: Sir Isaac Pitman & Sons, Ltd.
Johnston, E. (1980) Formal Penmanship: and other
papers (ed H. Child), New York: Taplinger Publishing
Company.
Johnston, E. (1986) Lessons in Formal Writing, H. Child
& H. Howes, (eds), Taplinger.
2. For more on Noordzij’s theories see:
Noordzij, G. (2005) The stroke: theory of writing, Eng-
lish translation, Hyphen.
Noordzij, G. (2000) Letterletter, Hartley & Marks.
3. For more on Blokland’s theories see:
Blokland, F.E. (2011) Harmonics, Patterns, and Dynamics
in Formal Typographic Representations of the Latin Script,
Thesis (PhD), Leiden University.
Golden section rectangle
Golden section rectangle
5
Stroke and
contrast
in history
Whatever one’s position on the question, the general understanding is that
the typefaces produced at the dawn of printing were all copies of contempo-
rary manuscripts, as that was the only lettering style known to the reader at
the time. Although the scribe culture of copying one book at the time eventu-
ally died out, defeated by the rapidly growing print industry, printing did not
completely replace handwriting. A substantial amount of non-printable mate-
rial continued to flourish in offices and private correspondence. The lettering
style of these messages was dominated by the writing masters of the day,
who published manuals instructing the public on how to write the various
hands of formal penmanship in vogue. The work of these writing masters had
a strong influence on the typographical development in the th and th
centuries.
Figure 5.2. Perpetua by Eric Gill. The
typeface Perpetua is based on Eric Gill’s
drawings with ‘brush and ink’; these were
taken over by Monotype, which was then
responsible for cutting the punches.
Illustrated with Perpetua.
At the turn of the 20th century, a common belief among scholars was that
many ancient scripts were drawn rather than written. This notion was forever
disproved by Edward Johnston’s demonstration that all these writing hands
could be created by a broad-nib pen held at various angles to the paper. As
discussed earlier, Johnston strongly believed in practising calligraphy as a
method for understanding letter construction. Others soon followed in his
footsteps. Among them was the Dutchman Jan van Krimpen (-),
who used the movement of the pen as an underlying force in his type design
work1. However, slightly less dogmatic than Johnston, Van Krimpen empha-
sised that the pen should only serve as an aid to the designer and never be
allowed to dominate the shaping of the type. Around the same time, the
American type designer Frederic W. Goudy (-) advocated the oppo-
site point of view, arguing against any inspiration from the pen, as he saw no
reason to “carry the qualities inherent in pen forms into letters produced by other
methods and for other purposes”2. This is a somewhat surprising statement
coming from Goudy, since many of his own types show a strong influence of
the pen stroke.
The discussion of whether typefaces exist on their own merits – separate
from any possible influence of the pen – or whether the pen is a leading force
in the shaping of all typographical styles continues to be a matter for discus-
sion among designers and scholars to this day.
Figure 5.1. Frederic W. Goudy and the
influence of the pen. Although Goudy
argued against the influence of the pen in
type design, the pen’s influence is in fact
evident in many of Goudy’s own designs.
Demonstrated with a specimen from ,
showing some of his work.
48 49
Figure 5.3. Jenson and Griffo. The cross-
bar of the Jenson ‘e’ has a diagonal angle
like a pen’s backstroke, and the lowercase
serifs finish to the right in a subtle upward
stroke in varying forms. In Griffo’s type,
on the other hand, the crossbar of the ‘e’
is horizontal, and his lowercase serifs are
uniform with a flat base similar in style to
his uppercase serifs. It appears that Jenson
was aiming at simulating the variations of
the penned letter, while Griffo’s Aldine types
aimed at a more uniform overall look.
Francesco Griffo for the Aldine Press, Hypnerotomachia Poliphili,
Nicolaus Jenson, EUSEBIUS, De Evangelica Praeparatione,
The Old Style stroke
The first printers were up against scholars who found books produced by
scribes to be far superior to any printed matter. To boost their business, print-
ers were therefore led to imitate the handwritten letters of the manuscripts.
The printer Nicholas Jenson (-) was one of the first to succeed
in printing with Latin type. As seen in his famous work from (Fig. .),
there is a strong reference to the pen in the modulated round forms. When
Bruce Rogers created his Jenson revival Centaur in , the shapes of the
characters were based on Rogers’ drawings of new letter shapes superimposed
on Jenson’s enlarged characters. Rogers later described how he had used a
broad-nib pen to write rapidly over the lowercase characters, while the upper-
case letters had required more careful drawing. In a letter written to Daniel
Berkeley Updike, Rogers explained his discoveries: “It proved to my own satisfac-
tion, at least, that the lower-case (with the exception of the s) of Jenson, was cut
directly from a MS hand — and not drawn — as the caps of course were”4.
The humanist book hands also inspired Aldus Manutius of the Aldine Press,
who in turn had a profound influence on future typefaces. His punchcutter
was Francesco Griffo (born ). Griffo apparently had a bad temper; the last
known record of his life is a charge of killing his son-in-law with an iron bar in
. Before this happened, Griffo had not only taken the roman fonts of the
Aldine Press in a more typographical direction (Fig. .) but had also produced
the first set of italic types ever applied in printing (Fig. .).
While these first printers of the Latin alphabet clearly took most of their
references from the hand-written manuscripts, the following generation
were less influenced by this connection. In an epistle from , typefounder
Claude Garamond (ca. –) explained how an acquaintance of his had
advised him to create type based on the italics of the Aldine Press, and so he
concluded, “What then was there to delay me after the encouragement, advice,
and help of such a generous friend? So I engraved italic types after the model of
the Aldine, with what success others will judge”5. The Aldine italic was based on
calligraphy if anything, yet in the text Garamond refers solely to copying the
Figure 5.6. Francesco Griffo revival.
The typeface Bembo, first published by
Monotype in 1929, is a revival of Griffo’s
type as shown in a book by humanist scholar
Pietro Bembo, published by Aldus Manutius
in 1495.
Figure 5.5. Jenson revival (two).
FB Hightower, created by Tobias Frere-Jones.
Figure 5.4. Jenson revival (one).
Adobe Jenson, created by Robert Slimbach.
Figure 5.7. Jenson revival (three). The
typeface Centaur is based on calligraphic
drawings by Bruce Rogers, created on top
of enlarged characters of Jenson’s type.
50 51
Figure 5.9. Revival of the Dutch tradition.
The typeface DTL Elzevir of the Dutch Type
Library is based on a collection of types attrib-
uted to the 17th century Dutch punchcutter
Christoffel van Dijck.
Figure 5.10. The dot of the ‘i’. In the early
years of printing, a common adaptation
from the calligraphic hand was to place the
dot of the ‘i’ slightly ahead of the stem.
Illustrated with books by Nicolas Jenson (top),
and Aldus Manutius (bottom).
fonts and makes no indication of other influences. Another concurrent exam-
ple is seen in a text published by Christophe Plantin in . Here the editor
(G.) and a printer (E.), presumed to be one of the brothers Robert or Charles
Estiennes, have the following dialogue:
“G. Well then, what are the principal parts of your art?
E. They are the types, the form or assemblage of them, and the press.
G. Is there anything else?
E. I am leaving out what we have in common with the writing master, like paper
and ink—although our ink is not like theirs”6.
If the norms of the time were that founders would study hand-written letters,
it would most likely have been mentioned in this context. However, the issue
is somewhat ambiguous, since Garamond is known to have based his Greek
types on a calligrapher’s handwriting. As Garamond was an educated crafts-
man, one would assume that he was also an experienced reader and writer.
Since the broad-nib pen was the writing tool of the time, writing with a pen
must have been a natural part of his daily life, and so its influence on his and
other contemporary type-founders’ work may have appeared so natural that
they saw no reason to mention it.
Figure 5.8. Garamond revival. The typeface
Garamond Premier Pro by Robert Slimbach
is based on Slimbach’s study of punches, by
the Frenchmen Claude Garamond and Robert
Granjon, found at the Plantin-Moretus Museum
in Antwerp, Belgium.
52 53
Romain du Roi
The typeface known as Romain du Roi is viewed as a milestone in the his-
tory of type design. With a long production period ranging from the years of
to , its constructed layout had a profound influence on the later
Transitional work of Baskerville and Fournier and also on the Didone faces by
Bodoni and Didot.
The typeface was commissioned by King Louis of France and created
by a committee appointed by the Académie des Sciences. The members of the
committee spent a significant amount of time researching manuscripts, cop-
perplate engravings, and printed books of previous generations. They further
studied contemporary examples by writing masters that practiced a lettering
style with thinner strokes and sharper serifs (Fig. .). These writing hands
were significantly ahead of the printing types of their time, which still relied
on the more than -year-old models of Garamond and his contemporaries.
Their study influenced the look of the new typeface and led to a style previ-
ously unseen in print.
The prominent French typefounder Pierre Simon Fournier later expressed
a rather critical view of the committee’s approach, which he saw as a rigid
system without room for optical corrections. In his Manuel Typographique from
-, Fournier asked rhetorically, “Are as many squares needed to make an
O, which is round, and so many circles to make other letters which are square? And
would not an artist have been any longer allowed to vary the shapes of letters,
both to height and width, to make minute graduations of size, as I have done,
as will be seen in the volume of specimens? What has been the fate of these
absurd rules? The fact is that the models given, especially for italics, are so ugly
and faulty that they denounce at once the cramping effect of all these circles
and squares in which they are enchained. Genius knows neither rule nor compass,
save in mechanical work”7. The statement is rather amusing, considering that
Fournier’s own work was so highly inspired by the Romain du Roi. Perhaps that
was the real reason for his criticism? It is not an unreasonable assumption
that the strong Romain du Roi references in Fournier’s typefaces caused him
to distance himself from the work of the committee. Yet, a key factor that
Fournier seems to have missed is that Romain du Roi never set out to dictate
a mathematical approach. In the documentation material from the project8,
the committee emphasised that they used their “eyes as judges”, and that
they had no intention of imposing the exact measurements of the large-scale
copperplate engravings onto punchcutting.
Figure 5.11. Calligraphic style inspiring
Romain du Roi. This handwritten book by
Nicholas Jarry (ca. 1610–1665) represents
the contemporary writing style that laid the
groundwork for the Romain du Roi design.
Note the x-height serif on the italic ‘h’ and
‘l’. The Romain du Roi committee adopted
this feature in the roman lowercase ‘l’.
L’office de la Vierge á mantines of 1661.
© Victoria and Albert Museum, London.
Figure 5.13. Pierre Simon Fournier.
Specimens from Fournier’s Caractères of 1742.
Figure 5.12. The Romain du Roi plates.
The proposed characters were presented
in a series of copperplate engravings with
geometric guidelines, developed by the com-
mittee as the model for Philippe Grandjean
(d. 1714) in his creation of the final punches.
While the ascending terminals of Old Style
typefaces ended in a diagonal cut inspired
by the broad-nib pen stroke, the strokes of
Romain du Roi had a vertical appearance.
The ascenders of the lowercase letters
terminated horizontally with a serif on both
sides. The axes of the round characters were
vertical, and the lowercase ‘b’ looks like a
reversed ‘d’.
54 55
Figure 5.14. Baskerville Original Pro.
A Baskerville revival published by Storm Type
in 2010. The typeface has two versions: one
that follows the original fonts with a small
x-height and high stroke contrast (120) and
one that is adjusted to modern needs with a
larger x-height and less stroke contrast (10).
The Baskerville stroke
The former writing master John Baskerville (-) decided late in life
that he wanted to be a printer, but he did not want to be a regular printer
– he wanted complete control of the ink, the paper and the typefaces of
his productions. By combining the readability of the Old Style faces with an
elegance and sharpness inspired by the copperplate engravings of the neo-
classical writing hands, Baskerville took his type to a level that is still a source
of inspiration for many present-day type designers.
Like the first printers, he transferred a calligraphic writing style to the
medium of punchcutting. When – after years of hard work in printing – he
finally felt ready to retire, Baskerville proposed the sale of his types and print-
ing equipment to the president of the Académie des Sciences in Paris. In this
letter he emphasised the originality of his own type by stating, “You will at a
Glance observe, that my Letters are not (one of them) copyed from any other; but
are wrought from my own ideas only”9. At the time, the custom of the English
Figure 5.15. The writing hand that inspired
Baskerville. Around 1715, the English writing-
master George Shelley published the book
Alphabets in all Hands, which demonstrated
examples of the new writing hand.
56 57
Figure 5.16. The Baskerville Specimen.
Baskerville’s work of 1757
printing industry was to create beautiful fonts in the spirit of the old punch-
cutters rather than developing their own concepts and ideas. So the fact that
Baskerville emphasised this quality in his sales pitch is rather extraordinary.
Yet, a visual inspection of some of the contemporary copperplate engrav-
ings shows that perhaps Baskerville was not quite as innovative as he himself
suggested. Baskerville’s type was less radical in the stroke contrast than the
penmen’s writing (Fig. .). To a certain degree, he adhered to the norms of
the printing industry by maintaining a level of contrast similar to the works
of his contemporary William Caslon (-), whose foundry at the time
was still producing types of the historical tradition today known as Old Style.
Baskerville was nonetheless moving away from the deep-rooted Old Style
types of his peers. The style was not new, however; he was simply follow-
ing the fashion of the popular writing hands. The esteemed type historian
Beatrice Warde later described the matter as follows: “This, then, was what we
owe to Baskerville: that at a time when abstract rules still were applied to works
of beauty—due to over-respectful study since the renaissance of didactic Roman
text-books on Greek art forms—he was sensible enough to base his design on
the living pen-form which will always remain as a check and inspiration to type
founders, rather than to following either the sterile pronouncements of archæolo-
gists on the one hand or the traditional rules of the punch-cutter on the other”10.
Printer and scholar Daniel Berkeley Updike (-) was not equally com-
plimentary. In his influential book Printing Types of , Updike argued that
in Baskerville’s type designs, “the hand of the writing-master betrayed itself, in
making them too even, too perfect, too ‘genteel’”11, a view that was further sup-
ported by type designer Frederic W. Goudy (-), who found the con-
trast between the thick and thin strokes too pronounced; he felt that “they
properly belong to plate engraving rather than to metal types”12, and, he argued,
they make the page appear restless and spotty.
In spite of this criticism, the Baskerville style is still highly popular.
Although the widely applied Vox-ATypI classification system of typefaces defines
the works of Baskerville as Transitional – between Old Style and Didone faces
– there can be no question that his work stands strongly on its own merits.
58 59
significant influence on the new letter shapes of the Didone types. That being
said, it is also evident that neither Bodoni nor the Didots found their inspira-
tion merely by following the work of formal penmanship. An examination of
their work shows a constructed influence on the forms, where a clear aim is
to create harmonious letters of high regularity and symmetry. They enjoyed
showing off their skills by producing prints with extremely thin hairlines, a
feature their predecessors had struggled to implement.
Figure .. The later Didot style. The
typeface Ambroise by Jean François Porchez
is a contemporary interpretation of the later
Didot types. Note the constructed elements
of the lowercase ‘g’.
Figure 5.20. Bodoni type. A page from
Giambattista Bodoni’s famous specimen
book Manuale Tipografico, finished by his
wife after his death and published in 1818.
Figure 5.18. The pen and its relation to the
fonts. The Copperplate engravings based on
the pointed-nib pen. Illustrated with plates from
George Bickham’s The Universal Penman of .
Figure 5.17. Giambattista Bodoni. Bodoni
was portrayed with a feather pen in his
Manuale Tipografico of 1818.
The Didone stroke
Combining the names of the Didot family in France and Bodoni in Italy, the
style of typefaces referred to as Didone has often been said to be the first to
finally break off the connection with the pen. This is not, however, completely
true. When printer and typefounder Giambattista Bodoni decided on a portrait
of himself for his Manuale Tipografico of , his choice of prop did not fall on
a specific drawing device, a mould or any of his punchcutting equipment; the
most important tool for him, and the one he chose to be immortalised with,
was a feather pen (Fig. .). As already discussed, the many new writing hands
had a strong influence on both the development of the Romain du Roi fonts
and the works of Baskerville; this source of inspiration was shared by Bodoni,
who is known to have been a big admirer of Baskerville. In one of his prefaces
Bodoni wrote of “the beautiful contrast as between light and shade which comes
naturally from any writing done with a well-cut pen held properly in the hand”13.
There is no doubt that the formal writing style practised at the time had a
60 61
The geometric stroke
A careful visual inspection of the development of the geometric sans serifs
shows a conceptual progression which initially adhered rigidly to a set of pre-
defined rules before gradually moving to a more pragmatic set of guidelines.
This development added nuance to the concept and enabled designers to cre-
ate types that appear geometric to the casual observer while still meeting the
needs of the human perceptual system.
The Bauhaus and the New Typography movements of the s believed
in type that was driven by pure functionality, where all ‘superfluous elements’
were eliminated; they viewed sans serif typefaces as the proper representa-
tives of this concept. These modernists praised the work of the engineers as
being non-artistic and function-driven, an approach that matched their own
aspiration of defining a new universal set of laws and an ideology that were
to be isolated from previous conventions and long-established traditions.
Several designers attempted to design systems along these guidelines,
among them the Bauhaus teacher Herbert Bayer (-) who devel-
oped his Universal alphabet around (Fig. .). Since type was produced
mechanically, Bayer argued against any influence from the calligraphic herit-
age. Instead, he argued in favour of letterforms with a monoline stroke width
based on perfect circles and straight lines, with as many identical letter
features as possible. None of the Bauhaus designers had a background in type
Figure 5.24. Absence of optical com-
pensation in Bauhaus lettering. With
no form of optical compensation in the
stroke weight and junctions, the geometric
letters drawn at the Bauhaus school were
conceptually stronger than their function-
ality. Illustrated with sans serif letters by
Joost Schmidt from .
Figure 5.23. Optical stroke compensation.
To make vertical and horizontal strokes
appear even in width, the vertical strokes
must be heavier than the horizontal.
Demonstrated with a mathematically accurate
circle at top, compared with an optically cor-
rected circle at the bottom.
Figure 5.21. The revival Ingeborg.
In the typeface family Ingeborg by Michael
Hochleitner, the text weights refer to the
Didone tradition, while the display weights
draw on the 19th-century lettering tradition.
Figure 15.22. A revival of the late th-
century Scottish tradition. Nick Shinn’s
typeface Scotch Modern is based on designs
that Shinn attributes to the typefounder
George Bruce. The work is located in a publi-
cation published in Albany in 1873.
62 63
Simultaneously with the German trend of geometric sans serif faces, several
English designers were working on low-contrast sans serif faces, starting with
Edward Johnston’s Sans for the London Underground, and followed by Eric
Gill’s Gill Sans. While the German modernists were keen on geometry, the
English traditionalists based their sans serifs on the humanist Old Style skel-
eton, changing the stroke contrast and eliminating the serifs.
The humanist sans serif has proven highly functional, with its built-in ref-
erences to the broad-nib pen and the simplicity of its detailing. The geometric
sans serif, on the other hand, has developed into a range of different styles,
most noticeable among them the works of the Dutch Functionalists of the
s, s, and s, which practise the grids and modular typefaces. Rather
than aiming for legibility, these types were seen as tools for sending a visual
message, like any other graphic element. To this day, the geometric sans serif
continues to be a highly popular style for creating bespoke typefaces for cor-
porate identity design. Due to its more recent arrival on the scene, it is subject
to fewer rules and has access to a larger playground, which leaves the designer
with more room for imbuing the type with personality and style.
Figure 5.26. The geometric Futura and
the humanistic Gill Sans. Like his modern-
ist contemporaries, designer Paul Renner
wanted to leave the influence of the pen
out of the letter shapes of his typeface
Futura, from 1927 (top), arguing that while
handwriting is based on a left-to-right
movement, the manufacturing method of
printing type is based on one impression
from above, which should be reflected in
the script of the future. In England, Eric
Gill was a contemporary of Edward John-
ston’s and strongly influenced by the pen
in his work. Although the basic structure
of Gill Sans (bottom) originates in the
humanist writing hand, the typeface has a
range of features that are similar to what
is seen in the German geometric typefaces
of the time.
Futura
Gill Sans
design, something that influenced their work, which tended to be created
strictly with a ruler and compass without any optical shape adjustment. A key
parameter for these letterforms was the uniformity of the stroke. The absence
of visual compensation between verticals and horizontals gave heavy-weight
letters a rather crude appearance.
The modernist designer Jan Tschichold, who in wrote the epoch-
making book Die Neue Typographie, which laid out the rules for the New
Typography movement, was trained in calligraphy. Tschichold’s background
gave him an eye for the details of letter shapes. Yet in spite of this, he
argued against the influence of the humanistic stroke on letters, since at
this point in life he advocated a rather radical practice in pursuit of a system-
atic approach to design. However, as evident in Tschichold’s suggestion for a
geometric sans serif typeface (Fig. .), Tschichold did demonstrate a concern
for optical compensation that was absent from the works of the Bauhaus
teachers. Figure 5.25. Sans serif letters by Jan
Tschichold. Although he was driven by a
fascination with defining a set of strict
universal laws to be followed by all design-
ers, Tschichold nevertheless used optical
compensation in his letter shapes, as dem-
onstrated in the narrowing of the round
strokes in the junctions14.
64 65
Figure 5.29. Bespoke geometric type for
. A collaboration between Studio 8 and
Dalton Maag, this bespoke typeface for the
Central School of Speech & Drama simulates
neon-lit signs from old cinemas and thea-
tres. Due to the inline feature, the solid bold
weight needed to be redrawn with far less
contrast between the vertical and horizontal
strokes than is normally seen when aiming for
an optical compensation of shapes.
Figure 5.30. Eurostile Next by Akira
Kobayashi and Sandra Winter. The aim of
this redesign was to get back some of the
spirit of softness that had been lost in
earlier digital versions of Aldo Novarese’s
original Eurostile.
Figure 5.31. The typeface Forza. A squared
geometric sans serif with personality, created
by Hoefler & Frere-Jones.
Fogerty Solid
Fogerty Inline
Figure 5.28. The typeface Avenir Next.
When type designer Adrian Frutiger in 1988
was looking back at the previous years of
sans serif faces, he found that the market
needed a more contemporary version of
the constructivist sans serif. The result was
the typeface Avenir. Frutiger was interested
in creating shapes that not only appeared
geometric but were in fact also optically ad-
justed. Yet in contrast to many of the former
designers, Frutiger wanted Avenir to have a
more humanistic appearance. This is evident
mainly in the lowercase letters of ‘a’, ‘f’ and
‘t’, which have more conservative shapes
but still work in the company of the other,
constructed lowercase letters. Furthermore,
the strokes in the junctions of the bold
weights are narrowed in a way that removes
the monoline element from the design. With
these adjustments, Frutiger has managed to
create a typeface based on constructional
principles that also works in running text.
In collaboration with Adrian Frutiger, Akira
Kobayashi developed the updated typeface
Avenir Next at Linotype in , shown here.
Figure 5.27. Dutch functionalism. The
geometric lettering by Wim Crouwel often
plays with the boundaries of legibility. Here
seen in the letters designed for the cover of an
artist catalogue from .
66 67
Figure 5.32. A geometric sans serif as
bespoke typeface in corporate design.
Geometric typefaces are rather popular
today in corporate identities. There appears
to be a consensus that the style represents
values that many modern organisations
would like to be associated with. Here illus-
trated with an exhibition identity for ‘Ergo-
nomics — Real Design’ at the London Design
Museum created by A/SW/HK in -.
Figure 5.33. Geometric typeface with a
horizontal emphasis. A bespoke typeface for
BBC One, created by Fontsmith, follows in the
tradition of Futura and Avenir. Yet, its vertical
terminals have a slight slope at the bottom,
leading the eye along to the next character.
1. Krimpen, J.v. (1930) ‘Typography in Holland’, The
Fleuron, vol.7, pp.1-26.
Krimpen, J.v. (1959) A Perspective on Type and Typogra-
phy, The Stinehour Press.
2. Goudy, F.W. (1942) The Alphabet: and elements of let-
tering, Berkley: University of California Press, p.81.
3. Rogers, B. (1949) The Centaur Types, Chicago:
October House.
4. Lawson, A. (2002) Anatomy of a Typeface, Boston:
David R. Godine, p.67.
5. Johnson, A.F. & Morison, S. (1924) ‘The Chancery
Types of Italy and France’, The Fleuron, vol.3, pp.23-51,
p.49.
6. Plantin, C. (1964) Calligraphy & Printing in the
sixteenth century: Dialog attributed to Christopher Plantin,
R. Nash (ed), Antwerp: The Plantin-Moretus Museum,
pp.37+39.
7. Carter, H. (1930) Fournier on Typefounding: The Text
of the Manuel Typographique (1764-1766), Translated into
English, London: Soncino Press, pp.8-9.
8. Jammes, A. (1965) ‘Académisme et Typographie:
The making of the Romain du Roi’, Journal of the Printing
Historical Society, no.1, pp.71-95.
9. Straus, R. & Dent, R.K. (1907) John Baskerville: A
Memoir, London: University Press, p.105.
10. Warde, B. (1927) ‘The Baskerville Types, a
Critique’, The Monotype Recorder, vol.26(221), Sept-Oct,
pp.3-29, p.12.
11. Updike, D.B. (1937) Printing Types: Their History,
Forms, and Use, 2 volumes, first published in 1922, revised
edition 1937, Cambridge: Harvard University Press, p.115.
12. Goudy, F.W. (1940) Typologia: studies in type design
& type making, with comments on the invention of typog-
raphy, the first types, legibility, and fine printing, Berkeley:
University of California Press, p.132.
13. Lawson, A. (2002) Anatomy of a Typeface, Boston:
David R. Godine, p.200.
14. Photograph by Christopher Burke, and taken
from his book Active literature: Jan Tschichold and New
Typography (London: Hyphen Press, 2007). The image is
reproduced with permission from the Tschichold estate.
References
68
6
The individual
letter
Researchers Sanford
()
Bouma
()
Tinker
()
Sanford
()
Geyer
()
Bouma
()
Dockeray
()
Test methods distance distance short
exposure
short
exposure
short
exposure
short
exposure
parafoveal
vision
Typefaces Old Style
roman
Courier low contrast
Didone style
Old Style
roman
Tactype
Futura
Courier Old Style
roman
Misreading y > p
i > l
w > v
h > b
m > w
b > h
p > r
n > a
h > k
t > i
e > c
l > i
f > r
l > j
k > x
c > e
o > c
v > r
q > g
y > r
j > l
m > u
c > o
l > i
g > q
m > n
w > v
e > o
i > l
c > e
h > b
b > h
r > f
z > i
t > i
g > v
o > n
c > o
s > e
s > o
k > h
z > r
y > r
y > p
f > t
a > d
h > b
j > l
b > h
f > t
t > f
c > e
e > c
i > l
i > j
m > n
n > a
u > a
l > j
q > d
w > u
y > v
k > h
v > y
m > w
p > b
x > z
f > l
w > v
m > w
j > l
l > i
r > f
h > b n
i > l
l < j
y > v
i > j
o > e
t > i
b > h
i > t
e > c
f > i
t > l
k > h
v > w
j > i
w > a
y > p
z > x
q > o
e > o
f > l
b > h
e > a
i > l
c > r
z > x
t > l
f > j
t > i
p > n
a > o
o > a
s > n
y > v
l > i
f > r
o > e
q > g
j > l
z > r
c > i
l > j
l > i
s > a
g > q
c > e
b > h
n > m
z > a
e > a
c > o
k > h
s > e
z > e
h > b
k > b
x > a
r > f
r > t
z > r
o > e
u > n
t > i
i > l
e > m
a > n u s
b > h
c > e o
e > c o g s
f > l t i
g > s
h > b k
i > j l t
j > i l
l > j t f
n > m u
o > n
p > m
q > o
r > f t
s > g
t > f
u > n
v > y
w > v
x >y v
y > v
z >x
Figure 6.1. The most frequently misread
lowercase letters. Applying different test
methods and test materials, this table pre-
sents a collection of studies looking into
the internal legibility of lowercase letters.
In all of the studies, aside from Dockeray’s,
the highest-frequency errors are listed first.
In the Dockeray study, the errors are listed
in no particular order.
The saying often referred to by designers that “type is a beautiful group of
letters, not a group of beautiful letters” is a good indication of how the overall
feel of a typeface is in many ways more important than the single characters.
Discussing individual letterforms within a typeface is therefore a difficult task,
since changing any single design feature often influences a whole range of
other elements. Not all letters that seem legible in isolation work when form-
ing words. Different neighbouring characters can influence the performance of
a letter in different ways. A designer always has to work on the whole and on
the individual elements simultaneously. As stated by Jan Tschichold, “A good
alphabet is like a harmonious group of people in which no one misbehaves”1.
Internal letter relations
Among earlier, traditional legibility researchers, a popular subject was to study
the relative legibility of letters by comparing the different characters within
the alphabet. A summary of this work2 found that certain distinguishing letter
features aid legibility more than others, and that lowercase letters such as
‘b’, ‘d’, ‘p’, ‘q’ and ‘k’ are among the most easily distinguishable due to their
descending or ascending elements and well-defined x-height features.
Figures . and . list a range of studies that focus on the misreading of
letters in different typefaces. A review of these studies reveals that differ-
ent typefaces produce different letter recognition errors. One example is the
study of the typeface Courier carried out by Bouma. Courier varies from other
common typefaces by being monospaced and having very large serifs (Fig. .).
Such features might result in some of the misreadings that are particular to
70 71
Figure 6.4. Commonly misread letter
groups. A close scrutiny of the data present-
ed in a number of scientific studies suggests
that these letter groups are the ones most
commonly confused. Illustrated with the type-
face EngelNewSans by Sofie Beier.
Figure 6.3. Test typefaces. The difference
found in letter shapes can influence test
results.
this specific typeface. Furthermore, Geyer’s study used the typeface Futura (Fig.
.), which features a ‘t’ created out of a straight vertical stem and no curve at
the bottom. This unusual shape probably increases the likelihood of the charac-
ter being misread as ‘l’ or ‘i’.
A review of the collected data shows a pattern of recurring misreadings in
the lowercase alphabet and identifies two main groups of problematic char-
acters. One group is composed of the x-height characters of standard width
based on a mixture of straight and curved lines (e-c-a-s-n-u-o); the other
group includes narrow letters with a single vertical stroke and small width
(i-j-l-t-f). The uppercase alphabet is also associated with several misreading
patterns, such as round shapes (, , , , ), diagonal shapes (, , , , ,
), vertical stem shapes (, , , ), shapes with mixed horizontal and vertical
strokes (, , , , , ) and shapes with two vertical stems (, , ).
Researchers Townsend
condition
()
Banister
table II
()
Tinker
()
Loomis
()
Phillips
et al.
()
van der
Heijden
()
Fisher et al.
table
()
Test methods short
exposure
short
exposure
short
exposure
blurring by
diffusion
distance short
exposure
short
exposure
Typefaces sans serif sans serif
Green’s
type
low contrast
Didone
style
Helvetica
Extra Light
Helvetica sans serif Futura
medium
Misreadings Q > O
B > R
F > T
T > I
J > I
H > N
L > I
D > O
G > O
O > G
U > H
Q > G
E > L
I > L
O > Q
M > H
A > K
X > K
G > H
R > H
P > F
F > I
I > T
V > Y
W > H
I > J
G > C
B > N
M > N
B > N
M > N
J > T
Q > G
V > Y
Q > O
G > C
M > N
Y > V
O > D
R > N
W > N
C > G
T > I
H > N
X > K
K > N
W > M
H > M
F > P
K > R
B > D
Q > D
X > N
C > O
D > O
K > L
B > N
D > G
P > F
Q > O
P > F
H > B
F > P
R > B
C > G
X > N
N > S
K > R
D > B
N > X
Y > V
R > H
L > I
O > Q
T > I
Y > X
E > F
E > L
U > I
U > L
G > C
G > O
F > I
H > U
H > R
J > I
T > Y
C > L
B > R
T > F
W > M
Q > O
M > U
Q > G
H > N
Y > V
P > F
F > P
V > Y
G > O
B > D
C > G
M > W
E > C
D > O
Y > F
Y > T
G > Q
O > Q
R > N
M > H
B > S
W > M
S > B
R > H
X > K
Z > J
B > O
K > X
H > M
B > R
S > R
Q > C
Y > T
M > W
F > T
H > M
R > A
B > P
K > X
W > M
G > Q
Y > V
E > C
B > R
H > W
H > N
Q > O
Z > J
X > A
K > A
Q > G
D > P
O > D
P > F
T > I
Z > I
E > J
E > K
X > K
N > M
O > M
G > N
D > R
K > R
Q > O
X > Y
S > G
F > P
E > F
Q > G
V > Y
K > X
T > I
O > Q
W > K
O > G
M > H
B > G
L > I
R > P
W > R
N > K
W > X
O > C
O > D
I > T
B > R
W > M
P > R
B > S
X > K
P > F
Q > D
W > N
U > W
I > J
T > I
Q > O
J > I
X > K
F > I
D > O
L > I
V > Y
K > N
C > O
F > P
I > T
S > B
F > T
G > O
R > K
U > O
P > I
C > G
E > I
E > L
H > N
I > J
K > R
L > T
E > F
Q > G
Figure 6.2. The most frequently misread
uppercase letters. Applying different test
methods and test materials, this table
shows a collection of studies looking into
the internal legibility of uppercase letters.
Highest frequency of errors listed first.
Courier
Futura
72 73
Figure 6.7. The areas we use to identify
letters. In a study of uppercase and lower-
case letters set in Arial, researchers6 applied
a method of short exposure, where a mask
of randomly placed Gaussian blur areas were
superimposed on each character. By sub-
tracting the blur areas that led to confusion
from the blur areas that allowed identifica-
tion, the researchers were able to locate the
letter features that facilitate recognition.
The conclusion was that both the stroke
terminations and the horizontal strokes
are crucial for letter recognition. The data
further showed that participants focused
on the areas that separate confusable letter
pairs from one another (note c-e-o).
Illustrated here with a close reconstruction by
the author.
Figure 6.9. When serifs improve legibility.
A study by Beier & Larson5 (left) showed
that in distance viewing, a serif on top of
the stem will improve the legibility of the
letter ‘i’. Harris3 had similar findings (right).
In his study, the ‘i’ and ‘j’ of Baskerville
proved to be more legible than the two sans
serif faces tested. Serifs appear to empha-
sise the cap between the stem and the dot.
Yet the lower x-height of the Baskerville
typeface, which places the dot away from
the stem, is also likely to have played a role
in the higher legibility of these Baskerville
characters.
Spencer Baskerville Univers Gill Sans
Figure 6.8. The middle section of ‘a’ and ‘s’.
A look at the test results presented in Figure
6.7 shows that the middle of the letters ‘s’
and ‘a’ are essential for letter recognition.
Research by Beier & Larson5 confirms this
finding by demonstrating that this area in
the two letters generally benefits from being
more rounded than diagonal, and that this
may in fact be more important than whether
the letters have open or closed apertures.
Gill Sans
Scientific results
Figure 6.6. The narrow letters. To make
room for the internal space of the charac-
ters, research suggests that wider forms are
preferable to narrower forms. One should
nevertheless be aware of the possibility of
introducing new types of misreadings when
broadening normally narrow characters. Har-
ris3 found the misreading of ‘t’ for ‘c’ to be
more common in the rather broad ‘t’ of Gill
Sans than in the narrower Univers and Bask-
erville versions. In general, however, both ‘t’
and ‘f’ performed better in Gill Sans than in
Univers, which is likely due to the slimmer
appearance of the characters in Univers. This
finding is confirmed by more recent research5
showing that narrow letters such as ‘j’, ‘t’
and ‘l’ benefit from being slightly broadened.
Figure 6.5. The skeletons of ‘c’ and ‘e’.
Based on the findings of his 1973 legibility
study of Baskerville 169, Univers Medium
and Gill Sans Medium, J. Harris3 recom-
mended open counters in the lowercase
letters ‘c’ and ‘e’ to avoid confusion with
the letters ‘o’ and ‘a’.
He also found the ‘e-c’ confusion to be
more likely in Baskerville, which has a high
crossbar, than in Gill Sans, which has a
lower crossbar. This was borne out in a later
study4, which found that a lowercase ‘e’
with a high crossbar has more misreadings
than a lowercase ‘e’ with a crossbar at the
visual centre.
Univers
c e o s a
b
f t i j l
h p n
74 75
Figure 6.10. The Johnston ‘l’. For his Lon-
don Transport typeface, Edward Johnston
applied a historically inspired calligraphic tail
to the lowercase ‘l’, a somewhat uncommon
feature in type design at the time. Walter
Tracy later recognised Johnston’s intention
of differentiating the lowercase ‘l’ from the
uppercase ‘I’. However, he emphasised that
when placed in words, the character “was
so broad that the letter stood aloof from the
one that followed”7. When Tracy was asked
in 1974 to redesign Johnston Sans, he con-
densed the broad ‘l’ noticeably. Illustrated
with Johnston’s original design.
Figure 6.11. The Garamond ‘t’. In his later
days Jan Tschichold had great admiration
for the types by Claude Garamond (d. 1561).
Tschichold praised the tail of Garamond’s ‘t’
for enhancing legibility and emphasised that
the top of the ‘t’ “is right in appearing to be
short”8. He further criticised contemporary
faces for lengthening the ascending part of
the ‘t’, as he found the feature to exacerbate
the risk of a misreading for the lowercase
‘l’. Tschichold applied this feature in his own
interpretation of Garamond’s work in the
typeface Sabon. Illustrated with the updated
version Sabon Next by Jean François Porchez.
Designers’ ideas
Figure 6.12. The cut-off ‘g’. The cut-off
tail of the lowercase ‘g’ is said to stem from
the 1905 German introduction of a standard
baseline for all typefaces. These guidelines
followed the proportion of the Blackletter
alphabet, and so provided very little space
for the descending parts. Whatever the
origin, the feature is usable in signage type-
faces, since it allows for the two parameters
of heavy weight and large x-height to be pre-
sent simultaneously in one font. Illustrated by
the typefaces Copenhagen by Henrik Kubel and
the Danish sign that inspired the design.
76 77
Figure 6.16. Design for readers with learn-
ing difficulties. In the process of designing
the typeface FS Me, Fontsmith consulted
a group of people with mild learning dis-
abilities and asked for their opinion on aes-
thetics and ease of reading. The consultant
group preferred the one-storey ‘’, long
ascenders and descenders were judged to
extenuate the letter shapes, and the let-
ters ‘i’ and ‘j’ have large dots to add clarity.
The letter group of ‘b’, ‘d’, ‘p’ and ‘q’ is not
a simple rotation: Each letter has its own
distinguishable shape. “What was important
throughout this process was to keep in mind
that ‘accessible design’ does not have to
be bland. The focus group felt very strongly
that a warm and soft design with strong core
shapes was coherent, readable and person-
able”, says Jason Smith, creative director
of Fontsmith.
Figure 6.15. Broad and narrow ‘r’. The
typeface Facit by Tim Ahrens has both a
narrow and a broad version of the letter ‘r’.
To maintain a harmonious negative space,
the ‘r’ is narrow when placed in front of
letters such as ‘a’, ‘t’, and ‘f’. To avoid a
misreading between ‘rn’ and ‘m’, the ‘r’
is broader under such circumstances.
Futura: Different termination of ‘c’ and ‘e’.
From the top: Gill Sans, Univers, and Avenir.
All with similar terminations of ‘c’ and ‘e’.
Figure 6.14. Eric Gill on the round upper-
case letters. To minimise confusion between
the uppercase ‘O’ and ‘D’, Eric Gill10 argued
against applying a basic form of a rectangle
of rounded corners, as he noted that al-
though it may look like an ‘O’, there is little
to differentiate it from a ‘D’. He goes on to
wonder why it appears to be the preferred
fashion of lettering created by engineers on
trains and of the newspaper vendors of his
time, since none of these purposes is purely
decorative.
Figure 6.13. Futura ‘c’ and ‘e’. In most
typefaces, the terminals on the lower parts
of the letters ‘c’ and ‘e’ are similar in shape;
however, in the typeface Futura by Paul
Renner, the terminal is cut off diagonally
in the ‘e’ to create a closed counter and
vertically in the ‘c’ to create a fairly nar-
row shape. This decision was based on the
German language having many ‘ch’ and
‘ck’ letter combinations; thus by creating
vertical stroke endings to the letter ‘c’, the
characters could be tightly spaced, which
would minimise disrupting gaps in the word
pattern9.
78 79
Figure 6.19. The serif on the ‘i’. The dots
on the ‘i’ and ‘j’ are in danger of visu-
ally connecting with the stem and being
perceived as ascending characters. This risk
is more evident in sans serif than in serif
typefaces. Although not approved of by
some designers – who argue that serifs do
not belong on a sans serif typeface – several
newer sans serif typefaces have a slab serif
on ‘i’ and ‘j’. The phenomenon is most com-
mon in typefaces designed for electronic
media and signage, where optimal character
recognition is thought to be essential, and
where spacing is less subtle. The slab serif on
‘i’ and ‘j’ is often found in the sans serif faces
by Erik Spiekermann, who further shows a lik-
ing for the sans serif ‘l’ with a tail to the right.
Spencer RegularFF DIN RegularClarendon
Figure 6.18. The uppercase serifs of the
typeface Spencer. To avoid the misreading
of ‘D’ as ‘O’, the stroke of the bowl extends
to the left of the stem in the upper part of
the ‘D’. For the sake of harmony, this fea-
ture is repeated in other uppercase charac-
ters that have a horizontal upper stroke (B E
F P R). The potential problem of misreadings
between ‘R’ and ‘B’ is addressed by an arc
attached to the right leg of the ‘R’ – the
arc is further applied to the other similarly
shaped characters ‘A’, ‘K’ and ‘k’.
Figure 6.17. The Spencer typeface’s
lowercase serifs. Based on J. Harris’ experi-
mental studies3, which found that serifs on
the lowercase counters appear to reduce
visibility, the serifs in the typeface Spencer
designed by Sofie Beier are removed from
the inner counters. To further enhance the
differentiation of ‘b’ and ‘h’, the right stem
of characters ‘h’, ‘n’ and ‘m’ extend in an
arc to the right, which ideally should further
enhance the horizontal flow of the font.
Meta
Unit
Info
Officina
Figure 6.20. The apertures of ‘a’ and ‘e’.
One of the great designers of the 20th cen-
tury, William Addison Dwiggins, advocated
an open white area in the letter ‘a’. He
further stressed that the generous openings
within the characters ‘a’ and ‘e’ should be
extended11. Illustrated with Eldorado Text by
Font Bureau, based on Dwiggins’ work of 1953.
80 81
Figure 6.22. The diagonal crossbar of
‘e’. A diagonal crossbar in the lowercase
‘e’ produces a large eye while keeping the
counter open. It is a feature inspired by the
Italian Renaissance humanist typefaces and
favoured by Frederic W. Goudy in his work.
A similar crossbar was also applied in the
typeface Clearface by Morris Fuller Benton.
Clearface was designed for high legibility
based on character differentiation. How-
ever, if the typeface is applied in running
text, the diagonal crossbar may possibly
counteract the horizontal flow. Walter Tracy
generally disliked the slanted bar of the ‘e’,
especially in sans serif faces, finding that it
disturbs the stability of a word and creates
an undesirable impression of restlessness7.
Illustrated with specimens from the American
Type Founders Specimen Book of 1923.
Figure 6.21. Condensed narrow letters
in sans serifs. In both slab serif and sans
serif faces, the lowercase ‘f’ is traditionally
rather narrow in width. This is partly due
to their origination in sign writing, where
space tends to be more limited in width
than in height, and partly due to their time
of arrival in the printing industry, which
corresponded with the existing fashion of a
slim ‘f’ in serif faces. The emphasis on the
narrowness of already narrow characters is
particularly evident in Futura. Not only does
the typeface have a remarkably slim ‘f’, the
characters ‘j’ and ‘t’ have no tails at all.
The concern that these unusual letterforms
would result in incoherent word patterns
was raised by several of Paul Renner’s con-
temporaries, but Renner was convinced that
the constructed geometric base that the
typeface was based on would tie the many
individual features into a unity of forms9.
Didot HFT: Didone tradition
of condensed narrow letters.
New Rail Alphabet by
Henrik Kubel: The sans serif
tradition of condensed narrow
letters.
Futura: Taking narrowness
to the extreme.
Capsa by Dino dos Santos:
The Old Style tradition of
expanded narrow letters.
82 83
Figure 6.23. Different stroke endings.
To reflect an informal quality of the
type, the stroke endings of the signage
typeface Wayfarer by Jeremy Tankard are
cut off at various angles. Most noticeable
are the angular endings of the top and
bottom stems of ‘l’ and ‘q’, contrasting
the horizontal cuts of the other vertical
strokes, and the different angles of the
tail endings of ‘c’ and ‘e’.
Figure 6.24. Typeface designed for the
Dutch telephone books. In the typeface
Telefont List for KPN, Martin Majoor differ-
entiates the letter ‘f’ from ‘t’ by making it
a descending character.
1. Tschichold, J. (1985) Treasury of Alphabets and Letter-
ing, first pub. in 1956, Omega Books, p.19.
2. Tinker, M.A. (1964) Legibility of Print, U.S.A.: Iowa
State University Press.
3. Harris, J. (1973) ‘Confusions in letter recognition’,
Professional Printer, vol.17(2), pp.29-34.
4. Fox, D., Chaparro, B.S., & Merkle, E. (2007) ‘Examin-
ing Legibility of the Letter “e” and Number “0” Using
Classification Tree Analysis’ [online], Usability News,
vol.9(2), available from: http://www.surl.org.
5. Beier, S. & Larson, K. (2010) ‘Design Improvements
for Frequently Misrecognized Letters’, Information Design
Journal, vol.18(2), pp.118-137.
6. Fiset, D., Blais, C., Éthier-Majcher, Arguin, M., Bub,
D., & Gosselin, F. (2008) ‘Features for Identification of
Uppercase and Lowercase Letters’, Psychological Science,
vol.19(11), pp.1161-1168.
7. Tracy, W. (1986) Letters of Credit: a view of type
Design, London: Gordon Fraser, p.89.
8. Tschichold, J. (1969) ‘Of what values is tradition
in type design’, Typographic Opportunities in the Computer
Age, Prague: Typograpfia, pp.52-55, p.53.
9. Burke, C. (1998) Paul Renner: The Art of Typography,
New York: Princeton Architectural press.
10. Gill, E. (1936) An Essay on Typography, London:
Sheed & Ward, p.44.
11. Dwiggins, W.A. (1947) Mss. by WAD: Being a collec-
tion of the writings of Dwiggins on various subjects Some
critical, some philosophical, some whimsical, New York: The
Typophiles, p.49.
References
84
7
Type for
text sizes
the sooner the reader identifies the word1. Findings like these show that when
we read running text, we not only see what is in the focus of our attention,
we also use the parafoveal area to gain helpful hints about what to expect
further down the line.
When we read small text sizes, we perceive a large number of letters
simultaneously in the centre of our field of vision. If a typeface has a high
degree of internal irregularity, this diversity will be very prominent in small
sizes since many of these elements will be presented to the reader at once.
As a result of this, a high degree of internal variation is not very suitable for
typefaces designed for blocks of text. As explained in by typographic
scholar and printer Theodore Low De Vinne: “The beauty of text-types is in their
precision. That freedom of drawing which is permitted, and sometimes approved, in
the letters of a good penman, or in engraving, or in the types of job printer, is not
tolerated in the text-types of books, which must be precise”.
Figure 7.2. Regularity in text type. When
reading a typeface in small point sizes, the
eye sees more irregular features than when
the same font is viewed in larger sizes at
the same distance. The letters of the text
version (top) of the typeface FF Info by
Spiekermann and Schäfer have a higher
regularity than the signage version of the
typeface (bottom).
In continuous reading, the eyes move along the line in rapid saccadic
motions. Between the saccades, the eyes stop and pause in fixations. As we
know, the Latin alphabet is read horizontally from left to right; nonetheless,
several of the letters have a structure based on vertical strokes. Many type
designers have tried to address this problem by introducing features that
make it easier for the eyes to stay on the same line when moving from one
fixation to the next. Some of these typefaces have serifs that are longer on
the right side of the stems than on the left, letters that lean forward slightly,
or a stroke contrast designed to motivate a horizontal direction. Whether this
is in fact effective is yet to be determined.
The retina has two types of receptors – rods and cones. They serve
very different purposes. The fovea at the centre of the retina consists
almost entirely of cones. Moving away from the fovea, the number of cones
decreases, while the number of rods increases. While rods can detect move-
ment and are sensitive in low illumination situations, cones process details
and register sharpness.
This means that the further away from the fovea an object is, the more
difficult it is to identify, as we lose retinal resolution. The fovea is obviously
essential to the perception of letters, but research shows that the area just
outside the fovea – the parafoveal area – is also very important. In fact, the
length of the next word in a sentence influences the length of the next sac-
cade: Long words lead to longer saccades, while shorter words produce shorter
saccades. If a word presented on a screen is changed as soon as a saccade
moves toward the word, the greater the similarity between the visual patterns
of the original word and the replacement word (chancing ‘home’ to ‘borne’),
Figure 7.1. Foveal, parafoveal and peripheral
vision. The fovea covers about 2 degrees
around the fixation point; outside the fixation
point lies the parafoveal area, which covers
about 10 degrees. Everything outside this is
referred to as peripheral vision.
love jogging
love jogging
FF InfoText
FF InfoDisplay
foveal parafoveal peripheral
86 87
Beautiful printing is an educator,
the same as is any art.
T
values by reason of it.
The mind is always receptive in proportion
AS IT IS HELPED TO
comprehend the real meaning of the writer.
Figure 7.6. FF Balance and the horizontal
focus. Designed by Evert Bloemsma and
inspired by the typeface Antique Olive, this
typeface applies an inverted stress to the
characters, with the horizontal parts being
heavier than the vertical parts. According to
Bloemsma, the inverted stress helps lead the
eye along and compensates for the absence
of serifs.
Figure 7.5. FF Quadraat and the horizontal
focus. A different way of emphasising the
direction of reading is demonstrated in the
regular styles of Quadraat by Fred Smeijers.
Inspired by early manuscript letters, the
characters have a subtle slant in the reading
direction.
TYPOGRAPHY SENDS
knowledge abroad as heaven
sends the rain. One fructifies
the soil, the other man’s intelligence.
Antique Olive by Roger Excoffon
FF Balance
hamburgefontsiv
Figure 7.3. Adapting the restlessness of the
early type in Kingfisher. Based on the idea that
the early punchcutters incorporated a certain
level of irregularity in their fonts to improve
the reading experience and interest the eye,
Jeremy Tankard built in an illusion of a slope to
the verticals, and added a form of movement in
several of the lowercase letters by varying the
angle of the axis.
Figure 7.4. The horizontal focus of the
typeface Swift. With large serifs and hori-
zontal termination of the round strokes,
Gerard Unger creates a strong horizontal
emphasis in the typeface Swift. Pictured
above is Neue Swift, a revised version of
Swift with a larger character set. Published
by Linotype in 2009.
Expert craftsmen magically produce
a wonderful instrument, which reveals almost incredible
fni ears
88 89
Figure 7.11. Designed to save space.
Fitting many words into a single line of
text is often a priority of newspapers. The
typeface Capitolium News by Gerard Unger
is designed to meet this requirement in
headline setting. For text sizes, Unger
designed the typeface Gulliver.
Figure 7.10. Walter Tracy and his ideas on
proportions. Walter Tracy advocates an in-
ternal relationship where the x-height is in a
ratio of about six to ten compared with the
ascending characters. According to Tracy,
it is more important to keep the ascending
characters (‘b’, ‘d’, ‘h’, ‘k’ and ‘l’) long than
it is for the descending characters (‘g’, ‘j’,
‘p’, ‘q’ and ‘y’).
for setting small text sizes or for saving space on the printed page often have
a large x-height and short extenders, the argument in favour of the large
x-height suggests that most of the essential letter features are to be found
in this area anyway, and therefore an enlargement of these features would
enhance letter legibility. This view is supported by an informal reading test
carried out by the designer Hermann Zapf7 in connection with the develop-
ment of the typeface Edison. In a comparison of different typefaces on poorly
inked newspaper, the study found that types with a large x-height delivered
the best performance.
POLITENESS MAY BE
everyday life
Proportions
Typefounder Pierre Simon Fournier ( –) had an eye for systems, and in
his Manuel Typographique from he suggested a relationship of lowercase
proportions where the x-height is units (about %) and the ascender and
descender units each3 (Fig. .). In a similar line of thought, more recently
the type designer Sumner Stone4 has advocated a relatively low x-height,
arguing that tall ascenders tend to emphasise the distinctness of letter and
word shapes and thus enhance legibility. Referring to the phenomenon that
the upper half of lowercase letters is read more easily than the lower half, the
type historian Harry Carter (-)5 suggested a solution where descend-
ing elements were kept short, and ascending elements long. Walter Tracy also
emphasised that if the x-height becomes too big, the extenders vanish and
the individuality of characters is reduced6. However, typefaces created either
Figure 7.7. The x-height of Mrs Eaves. The
highly popular typeface Mrs Eaves, designed
by Zuzana Licko in 1996, is inspired by the
work of John Baskerville. To enhance the
openness and lightness of the Baskerville
fonts, Licko gave the lowercase letters wider
proportions; she then made the x-height
smaller to prevent the type from taking
up too much space. These proportions
make the type highly functional for shorter
paragraphs of text in larger sizes but less
functional for longer type settings. As a con-
sequence, in 2009 Licko released the larger
x-height typeface Mrs Eaves XL.
Figure 7.8. The different proportions
of Lexicon. The typeface Lexicon by Bram
de Does has two versions with different
extending lengths: one with exceptionally
short extenders, designed for the setting of
small text in dictionaries, and one with more
conventional length.
Figure 7.9. Fournier and letter proportions
Pierre Simon Fournier (1712-1768) recom-
mended an x-height of 3 units and extend-
ers that are 2 units each.
Capitolium News
Gulliver
90 91
Figure 7.13. Dwiggins’ M-formula in type.
William Addison Dwiggins implemen ted his
ideas of sharp forms in his type design, as
seen here in his typeface Caledonia (below),
calling the abrupt change in the counter a
‘calligraphic flick’. This sharpness of the in-
ner curve of the ‘h’ inspired Kent Lew in his
design of the typeface Whitman published
by the Font Bureau (right).
Figure 7.14. Sharp serifs and stroke endings.
By sharpening omnipresent features like
serifs and stroke endings, Thomas Gabriel
implemented the M-formula ideas in his type-
face Premiéra, designed for small text sizes.
The M-formula
Among other things, William Addison Dwiggins enjoyed creating marionettes.
In , while carving a head for a new puppet, he made an interesting dis-
covery. Dwiggins found that to successfully carry the facial expressions of a
young girl to observers at the back of the room, the otherwise soft features
should be carved as sharp edges (Fig. .). At a distance, these exagger-
ated features would appear just as gentle as they were originally intended.
Dwiggins later transferred this knowledge to his text typefaces by sharpening
the character edges. For him, the discovery was a way of tricking the eye into
seeing nonexistent curves in objects of reduced sizes while enhancing the
character’s features. Named after the marionettes, he called this observa-
tion his M-formula. In a discussion of Dwiggins’ theory, type designer Gerard
Unger supports the idea that the influence of distance and light on mari-
onettes can be transferred to small text sizes, yet he emphasises that the
technique will be most effective in newsprint at point sizes of seven or less8.
Based on the notion that text in small sizes and text viewed at a distance
have a similar visual angle, and since the M-formula originates in a situation
involving distance, one can assume that the approach would be useful when
applied to type designed for distance reading as well.
Figure 7.12. Dwiggins’ M-formula in
marionettes. William Addison Dwiggins’
illustration of the puppet, with the soft
fea tures to the left and the sharp fea-
tures to the right.
A. B.
92 93
Figure 7.17. The ink traps of Retina.
The typeface Retina, designed by Hoefler
& Frere-Jones for 5-point setting on the
Wall Street Journal stock pages, with ink
traps designed to fill in on the press.
Ink and printing
One way of preventing ink from blurring the letterforms in smaller sizes is to
open up the junctions in ink traps. An early example of this, where the outside
of the stroke is cut off at a straight angle, is demonstrated in the work of
the German punchcutter Johann Michael Fleischman, who was employed by
Enschedé from to (Fig. .).
The typeface Bell Centennial (Fig. .) is famous for its ink traps. Created
in by Matthew Carter and designed for the phone directories of AT&T,
these ink traps differ from Fleischman’s, as the edge of the strokes bends
inwards instead of being cut off at the side. The unusual shape originate in
trial pages printed under production conditions, which allowed Carter to study
the ink spread and make changes to the letterforms accordingly. As Carter
explains: “Taking weight out of the intersections of strokes enhanced the clarity
of complex forms. The fact that these compensations were more drastic than usual
was really a function of the way the type was set and printed”9.
Figure 7.16. Bell Centennial. Designed for
small text sizes in telephone books, Matthew
Carter opens up the junctions to avoid over-
inking in the typeface Bell Centennial.
the quick brown fox
1234567890
dog
jumps over a lazy
Figure 7.15. Johann Michael Fleischman.
Ink traps in an 8-point type from the 18th
century.
94 95
Figure 7.19. The different grades of the
typeface Mercury Text. Ink acts differently
depending on the absorption of the paper
on which it is printed. Designed to accom-
modate different reproduction technologies,
the text fonts from the Mercury typeface
family by Hoefler & Frere-Jones each have
four grades available. These offer lighter and
darker versions of a same master design,
within the same horizontal size.
The idea is to provide designers with suf-
ficient options to avoid letter shapes from
dissolving in low-quality printing and to
further ensure the right relationship be-
tween Regular and Bold weights.
Grade Grade Grade Grade
In , when Martin Majoor tested a new typeface for the Dutch telephone
books, he established that the spikes and ink traps of Bell Centennial were
no longer necessary due to the new printing techniques. A similar finding is
reported by Bruno Maag10 when designing the typeface for the BT directo-
ries using both coldset and hotset printing (Fig. .). This reported quality
improvement does not appear to be equally strong in all printing scenarios.
Thus, when designing the typeface Retina for the stock page of the Wall Street
Journal, Hoefler & Frere-Jones found that small types in newsprint still need
strong ink traps.
Figure 7.18. Different printing techniques
of the BT directories. The typeface devel-
oped for the BT directories by Bruno Maag
is designed for printing under two different
techniques (shown here in development
trials). In hotset printing (left) the pages are
dried immediately after printing, while the
coldset printing (right) dries on its own. This
results in the ink of coldset printing bleed-
ing slightly more than the other, and so the
type has a heavier appearance.
96 97
Type for screen
The resolution of the screen controls the presentation of the font. Large
sizes will have more pixels per em (ppem), while small sizes will have fewer.
Consequently, screen types at small text sizes have to conform to a grid con-
sisting of a limited number of pixels.
In a look at what has been done to meet the requirements of small text
size screen presentation, the work carried out by Matthew Carter in the s
is famous for the reversed process. Instead of designing the outline of the
fonts first and then placing the pixels within the outline, Carter designed the
computer screen fonts for the Verdana, Tahoma, Nina and Georgia series the
other way around. After first creating a bitmap version of a particular target
size, he then designed the outline shape of the characters around the bit-
maps (Fig. .). The method is similar to that applied today by David Berlow
of Font Bureau12 when developing new web fonts for the Reading Edge series.
Here the fonts are designed on a grid further divided into a unit
em-square, which allows Berlow to ensure that all fonts regardless of hinting,
Figure 7.23. Before ClearType. Hinting of
a small-size italic ‘h’ letter, from the typeface
Georgia designed by Matthew Carter11.
Figure 7.24. The ClearType Technology.
A small-size italic letter ‘h’ from the typeface
Constantia by John Hudson. The image shows
the subpixels that make up the letter on
the screen at this size. Constantia has been
outfitted with ClearType hinting instructions
to take advantage of this subpixel grid11.
Figure 7.20. Lightening the heavy weights
of Trilogy Egyptian. Inspired by the Victori-
an-era Egyptian typefaces, Jeremy Tankard
was interested in making the style less geo-
metrical than what is seen in many earlier
revivals. His approach was to open up the
junctions, a solution that not only provides
ink traps but also gives the heavy Egyptian
style a lighter feeling without reducing the
weight.
Figure 7.22. Opal and ink traps. In the
typeface Opal by Hannes von Döhren,
the ink traps act as an outline cutting
into the letter shape in the junctions.
Figure 7.21. Ink traps for personality.
The ink traps of the typeface FS Rufus by
Fontsmith not only improve legibility in small
point sizes but also extenuate the soft and
mild personality in the larger sizes.
98 99
Figure 7.27. FS Joey by Fontsmith. The
constructed typeface FS Joey was origi-
nally created for a video-on-demand online
service. It is designed to function both
onscreen and in print.
will render as intended, at least at this size. The fonts are furthermore
designed so that variations in overshoot and stem width are simplified out,
as these are features that would otherwise disappear when rendered below
ppem (Fig. .).
In Microsoft introduced the series of ClearType fonts designed for
the newly developed technology of subpixel rendering, which controls the
red, green, and blue () elements of each pixel (Fig. .). In doing so, the
aim was to optically enhance the resolution of the screen by controlling even
smaller units within the pixel. Unfortunately, this effect only works on the
vertical strokes; Microsoft therefore mixes subpix el rendering of the vertical
strokes with anti-aliasing (shades of grey) for all strokes. Based on this mix
of techniques, and to minimise the jagged diagonals, several of the design-
ers behind the ClearType fonts enhanced the square feeling of the types by
emphasising horizontal and vertical lines (Fig. .)13.
A visual inspection of various other fonts for the computer screen dem-
onstrates a general tendency toward generously spaced letters, high x-height
and a low stroke contrast. Furthermore, many current screen devices are not
sensitive to kerning. To keep an even flow in the word pattern, type design-
ers can benefit from revisiting some of the techniques applied by early th
century designers who were compelled to create kernless typefaces for the
Linotype line-caster machine.
Figure 7.26. Enhancing the horizontals
and verticals. Because subpixel render-
ing only works on the vertical strokes the
ClearType team focused on minimising the
area of transition between verticals and
horizontals. The team further found that
ClearType hinting and rendering motivates
triangular serifs. Illustrated with Constantia
by John Hudson (top), and Calibri by Luc(as)
de Groot (right).
hamburgefontsiv
ag
Figure 7.25. The Reading Edge fonts by
Font Bureau. The Reading Edge typeface
PoynterSerif RE has a larger x-height and
wider characters than the print version of
the typeface, Poynter Oldstyle.
PoynterSerif REPoynter Oldstyle
100 101
Figure 7.30. Optical sizes in sans serif
and serif faces. Due to their higher stroke
contrast, serif faces often benefit from
optical scaling in ways that would not help
sans serif faces noticeably. The super-family
Freight by Joshua Darden features both sans
and serif versions, yet only the serif versions
contain optically adjusted fonts (note the
M-formula features of the micro sizes).
Figure 7.29. Optical sizes in FF Clifford.
The typeface FF Clifford by Akira Kobayashi
is inspired by Alexander Wilson’s Long Primer
Roman and Joseph Fry & Son’s Pica Italic.
The family has three optical weights, based
on Kobayashi’s analysis of old metal type
specimens.
FF Clifford Six Roman
FF Clifford Six Italic
FF Clifford Nine Roman
FF Clifford Nine Italic
FF Clifford Eighteen Roman
FF Clifford Eighteen Italic
VVVVVVVVVVVVVVVV
VVVVVVVVVVVVVVVV
Freight Micro
Freight Text
Freight Display
Freight Big
Freight Sans
Scaling
In the early th century, the printing process was more or less similar to
the methods applied at the time of Gutenberg some years earlier. This
changed drastically with the advances of the Industrial Revolution. By the end
of the th century, nearly all stages of printing could be power-driven. The
influence of the general increase in the speed of printing was substantial. Of
all the new inventions, the most remarkable is the Benton pantograph engrav-
ing machine, developed in . This machine finally eliminated the necessity
for punchcutting in type design. It suddenly became possible to scale type
based on one master drawing alone.
In small print sizes and especially in low-quality printing, the letterforms
tend to melt away on paper. With manual punchcutting, each point size is cut
individually, which gives the gifted punchcutter the option of optically scaling
the fonts to suit the requirements of the paper, the ink, and the human eye.
To optically compensate type for small sizes, Walter Tracy14 suggested a gen-
eral widening of the characters and moving the baseline a little lower on the
body; this results in shorter descenders and larger x-height characters. Harry
Carter15 also advocated short ascenders, slightly heavier weights, low contrast,
and magnified strong serifs. He further stressed the importance of gener-
ous white space inside the letters, which was his reason for recommending
broader characters to preserve the balance between black and white.
Figure 7.28. Optical scaling by Bodoni.
Fonts from Bodoni’s Manuale Typografico
of 1818; for this purpose the letters have
been scaled to the same size, with the
smallest fonts at the top. Figure 7.31. Simplification of small point
sizes. The typeface Minuscule by Thomas
Huot-Marchand, has five different optical
sizes. The typeface is based on the theories
of French ophthalmologist Louis Émile
Javal, practising in the late 19th century,
who found that for very small sizes the
reader tends to mainly read the difference
between the letters. Huot-Marchand works
with an extreme simplification and differ-
entiation of the letters in the smaller point
size fonts, aiming at a typeface readable in
sizes as small as 2 points.
102 103
Figure 7.33. Weight and scale. The stroke
weight of a typeface will appear differently to
the reader depending on the size in which it is
presented; thus, a functional bold in 10 points
is not necessarily the same as a functional
bold in 300 points. Edward Johnston, with his
focus on calligraphy writing, found medium
weights more legible than heavier weights
and defined the optimal stroke weight to be
of about one fifth of the x-height16. Illustrated
with the typeface London by Henrik Kubel.
Micro Regular
Micro Italic
Micro Bold
Micor Bold Italic
Text Regular
Text Italic
Text Bold
Text Bold Italic
Display Regular
Display Italic
Display Bold
Display Bold Italic
Transitional Didone
eco
fja eco
fja
eco
fja
Figure 7.32. Applying several typeface
categories within one family. The typeface
Pyke by Sofie Beier is inspired by the work
of Giambattista Bodoni. The small sizes are
designed for running text. These sizes there-
fore have a horizontal emphasis, as do the
early Transitional style typefaces by Bodoni.
The Display sizes, on the other hand, are all
designed for reading a few words at a time;
therefore, the stronger Didone style features
may be present without sacrificing legibility.
abc
References
1. Rayner, K. (1978) ‘Foveal and parafoveal cues in
reading’, in: J. Requin, (ed), Attention and performance VII,
Hillsdale, NJ: Erlbaum.
Rayner, K. & McConkie, G.W. & Ehrlich, S. (1978) ‘Eye
movements and integrating information across fixa-
tions’, Journal of Experimental Psychology: Human Percep-
tion and Performance, vol.4, pp.529-544.
2. Vinne, T.L.D. (1900) The Practice of Typography: A
Treatise on the processes of type-making, the point system,
the names, sizes and styles of Printing Types, New York: The
Century Co., pp.11-12.
3. Carter, H. (1930) Fournier on Typefounding: the Text
of the Manuel Typographique (1764-1766), Translated into
English, London: Soncino Press.
4. Stone, S. (1989) ‘Hans Eduard Meier’s Syntax-
Antiqua’, in: C. Bigelow, P.H. Duensing & L. Gentry, (eds),
Fine Print On Type, London: Lund Humphries, pp.22-25.
5. Carter, H. (1937) ‘Optical scale in typefounding’,
Typography, vol.4, pp.2-6.
6. Tracy, W. (1986) Letters of Credit: a view of type
Design, London: Gordon Fraser.
7. Zapf, H. (1987) Herman Zapf & His Design Philosophy,
Chicago: Society of Typographic Arts.
8. Unger, G. (1981) ‘Experimental No. 223, a news-
paper typeface by W.A. Dwiggins’, Quaerendo, vol.11(4),
pp.302–324.
9. Information supplied through e-mail correspond-
ence between the author and Matthew Carter.
10. Information supplied through e-mail correspond-
ence between the author and Bruno Maag.
11. Berry, J.D. (2004) Now read this: The Microsoft
ClearType Font Collection, J.D. Berry & J. Hudson, (eds),
Microsoft Corporation, pp.8-9.
12. Information supplied through e-mail correspond-
ence between the author and David Berlow.
13. Berry, J.D. (2004) Now read this: The Microsoft
ClearType Font Collection, J.D. Berry & J. Hudson, (eds),
Microsoft Corporation.
14. Tracy, W. (1986) Letters of Credit: a view of type
Design, London: Gordon Fraser.
15. Carter, H. (1937) ‘Optical scale in typefounding’,
Typography, vol.4, pp.2-6.
16. Johnston, E. (1980) Formal Penmanship: and other
papers (ed. H. Child), New York: Taplinger Publishing
Company, p.49.
104
8
Type for
distance
viewing
Compensation for loss of details
The type historian Harry Carter argued that both serifs and high stroke
contrast impair the reading experience for typefaces viewed at a distance2.
The stone carver and lettering artist David Kindersley (-) disagreed
strongly with this view. Kindersley believed that heavy strokes of low contrast
in signage typefaces would obstruct the open counters and reduce legibility
when viewed from an acute angle3. Like others, Kindersley had observed that
both corners and characteristic parts of letters have a tendency to become
rounded and lose definition when viewed from afar. Kindersley’s solution in
the street sign lettering he designed (Fig. .) and later in his proposal for
the British road and motorway sign system was to apply serifs to the letters;
he found that the “serif reinforces the individual character of the letter exactly
where this loss is greatest”4.
So what is the best way to treat the corners of a signage typeface? An
anecdote related to the origin of slab serif typefaces is interesting in this
connection. Most historians5 seem to be puzzled by the fact that the slab
serif style was originally named Egyptian. One type historian6, however, points
to a possible logical explanation. The legend goes that during Napoleon’s
Egyptian campaign, the army communicated by placing stations at intervals
of every few miles. Their role was to display painted messages on large boards
Figure 8.1. The street sign typeface by
David Kindersley. David Kindersley advo-
cated the use of serifs on typefaces that
are to be viewed at a distance. Illustrated
with the typeface Kindersley Grand Arcade, a
free digital version recently produced by the
Kindersley workshop.
KINGSLAND ROAD
OLD STREET STATION
In a study of the difference in the rate of confusion between letters read at
distance in the foveal view and letters read up close in the peripheral view,
researchers1 found that curved letters such as ‘D’ and ‘O’ are more easily
confused in close-up peripheral viewing than they are in distance viewing.
Comparing this with the fact that different test methods generally produce
different results, the collective data suggest that our perception of let-
ter shapes is heavily influenced by the condition under which the letter is
presented. The lack of information we experience as the number of cones
decreases in the peripheral area of the retina may not be identical to the blur
we experience when we focus on something in the distance with all the cones
in the fovea in use (see more in Chapter ). However, in distance reading, large
type sizes will appear tiny, meaning that a letter that is two metres tall when
viewed at the proper distance will have the same visual angle as a -point
letter. Such situations cause some issues to be identical between type for
signage and type for micro text.
If, on the other hand, we look at large-size type viewed up close, few let-
ters appear in the foveal area at a time. A consequence of this is that readers
will find a higher differentiation level of the font acceptable in signs. Also, the
task of moving the eye along the text is not as big an issue when it comes to
short statements compared with what is involved in reading running text.
Figure 8.2. Round corners on FF InfoDis-
play. Erik Spiekermann and Ole Schäfer kept
the corners of the typeface FF Info rounded,
as they found that this feature on backlit
signage prevents light shatter and makes it
easier to create an even appearance. It also
saves considerable time when plotting char-
acters because the blade on the cutter does
not have to be lifted and turned degrees
but can continue to cut in one line. Illustrated
with a photo of a sign from Düsseldorf Airport
(top) and a digital rendering (bottom).
that could be read by telescope from the next station, which would then relay
the message to the following station, and so on. The letters used for this
task were apparently slab serif faces, which due to the heavily squared serifs
appeared to be more distinguishable at a distance.
Although this is a nice story, the anecdote is contradicted by the fact that
at the time of the first slab serif and sans serif type specimens, the slab serif
was not the only one called Egyptian. For a while, the designations Egyptian and
Antique were applied interchangeably to both sans serif and slab serif faces.
Figure 8.5. Round corners on M.O.L. In ,
when creating the backlit signage typeface
M.O.L. for the Amsterdam metro, Gerard
Unger based the typeface on the observa-
tion that light shining through an opening
of any shape always tends to form a circle.
Figure 8.3. Large junction by Engelhard.
Originating in the technical properties of
enamel signs, the Danish architect and
designer Knud V. Engelhard (-) had
a love for the large squared junctions in let-
ters such as ‘A’ and ‘M’.
Figure 8.4. Large junction by Linnemann.
To improve legibility by opening the inner
space of the letters and avoid cluttering
both on signs and in small print sizes, the
Danish type designer Bo Linnemann has
adopted Engelhard’s large junctions in many
of his typefaces.
AtB by Bo Linnemann & Elias Werner
Danske Headline by Bo Linnemann
Figure 8.8. The most legible? Test material
applied in a legibility investigation for
Heathrow Airport signage7. The typefaces
compared were as shown (from top)
Sign, Frutiger Bold, Frutiger Roman, Vialog,
and Stempel Garamond Italic, with Sign
shown in outline. The study found the nar-
row Vialog to deliver an inferior performance
compared to Sign and the two Frutiger
weights.
Figure 8.9. The weight of Cheltenham Old
Style. In a comparison of Cheltenham Old
Style (Regular), Bold and Bold Condensed,
Roethlein9 found Cheltenham Bold to be
more legible at a larger distance than either
the Regular or Bold Condensed, a finding
that not only suggests that wider typefaces
are more legible at a distance than narrow
ones but also indicates that bold weights
on non-reflective materials have superior
distance performance compared with regu-
lar weights.
Proportions
To facilitate larger point sizes and thus improve distance viewing, many
signage typefaces have a narrow width. This appears to be based on a mis-
conception. A legibility study7 of typefaces for signage at Heathrow Airport
found that the typeface Vialog appeared to be less legible than Frutiger Bold,
Frutiger Roman, and the typeface Sign (Fig. .). The fact that Vialog,
which is specifically designed for high legibility, performed so poorly is rather
interesting. Since the widest of the typefaces had the best performance and
the narrow one performed poorly, it is possible that the poor Vialog results
are caused by the width of the design. The findings are further confirmed by
a number of distance studies8 in which narrow typefaces all deliver the worst
performance, and the widest typefaces the best performance. This suggests
that legibility at a distance is not necessarily improved by applying larger
condensed faces. Furthermore, while both continuous text and road signage
are mostly read in a frontal position, other signs will often need to be read
at a more acute angle, which will cause condensed characters to appear even
narrower. Figure 8.6. Halation and the typeface
ClearviewHwy. Not all signs benefit from a
bold weight. Backlit and retroflective signs
tend to make type of a heavy weight bleed
out by creating an overglow effect on the
surface. A simulation of halation showing the
effects of overglow on the typeface
(previously applied on highway signs) vs. the
typeface Clearview by Terminal Design and
Meeker & Associates.
Figure 8.7. Junctions and the typeface
ClearviewHwy. To avoid the downfalls
of halation, the design team behind the
ClearviewHwy focused on creating open
junctions and large counters in the low-
ercase letters ‘a’, ‘e’, and ‘s’. Comparative
anatomy of the Series E-modified and
the typeface Clearview -W.
FHWA Series E(modiefied)
Clearview -W
FHWA Series E-modified
Clearview -W
FHWA Series D
1. Reich, L. N. & Bedell, H. E. (2000) ‘Relative legibil-
ity and confusions of letter acuity in the peripheral and
central retina’, Optometry and Vision Science, vol.77(5),
pp.270-275.
2. Carter, H. (1931) ‘Sanserif Types’, in: O. Simon,
(ed), The Curwen Press Miscellany, London: The Curwen
Press, pp.35-45.
3. Dreyfus, J. (1957) ‘David Kindersley’s contribution
to street lettering’, Penrose Annual, vol.51, pp.38-41.
4. Kindersley, D. (1960) ‘Motorway sign lettering’,
Traffic Engineering & Control, vol.2(8), pp.463-465, p.465.
5. Johnson, A.F. (1934) Type Designs: Their History and
Development, London: Grafton & Co., p.202.
Tracy, W. (1994) ‘Why Egyptian?’, Printing History,
31/32, pp.3-10.
6. Denman, F. (1955) The Shaping of our Alphabet, New
York: Alfred A. Knopf.
7. Waller, R. (2007) ‘Comparing typefaces for airport
signs’, Information Design Journal, 15(1), pp.1-15.
8. Forbes, T.W. and Holmes, R.S., (1939) ‘Legibility
Distances of Highway Destination Signs in Relation to
Letter Height, Letter Width and Reflectorization’, Pro-
ceedings of the Highway Research Board, vol.19, pp.321-355.
Zwahlen, H.T., Sunkara, M. & Schnell, T. (1995) ‘A
Review of Legibility Relationships Within the Context
of Textual Information Presentation’, Transportation
Research Board, no.1485, pp.61-70.
Schnell, T. (2008) Legibility optimization of uppercase
alphanumeric text for display messages in traffic applica-
tions, PhD Thesis (unpublished), College of Engineering
and Technology, Ohio University.
Garvey, M., Zineddin, A.Z. & Pietrucha, M.T. (2001)
‘Letter legibility for signs and other large format appli-
cations’, Proceedings of The Human Factors and Ergonomics
Society 45th Annual Meeting, pp.1443-1447.
9. Roethlein, B.E. (1912) ‘The relative legibility of
different faces of printed types’, American Journal of
Psychology, vol.23(1), pp.1-36.
About years ago, researcher Barbara Elizabeth Roethlein9 set out to
measure the legibility of a range of different typefaces in distance threshold
studies. Although it was not the main purpose of her study, the findings offer
useful information about typeface weight and proportions. In her list of The
Average Legibility of Various Faces (Fig. .), we see an almost even scale, where
the types with the largest x-height are the most legible, while those with a
smaller x-height are less legible. A further analysis of the findings suggests
that bold weights and low stroke contrast additionally enhance distance
visibility on the non-reflective surfaces applied in the experiment. The same
results would probably not be achieved if the letters were presented on high
brightness material. With white retroflective text on a dark background, under
those conditions heavy strokes will become too bright, bleed out, and visually
close up the inner counters of the characters.
References
Figure .. The average legibility of
various faces. This overview of Roethlein’s9
findings demonstrates that typefaces with
a large x-height, heavier weight, and broader
width perform better than others.
9
The capitals
Roman inscriptions
Traditionally, the uppercase letter style can be divided into two categories;
one refers to the proportions of the Roman inscriptional capitals, and the
other aims for an approximately even width between the letters. The English
printer Joseph Moxon is one of the earliest known sources to show a con-
cern for the legibility of the Latin alphabet. In Moxon stated: “we must
conclude that the Roman letters were Originally invented and contrived to be
made and consist of Circles, Arches of Circles, and straight Lines; and therefore
those Letters that have these Figures, either entire, or else properly mixt, so as
the Course and progress of the Pen may best admit, may deserve the name of true
Shape, rather than those that have not”.
Figure 9.3. The variations of the Roman
letters. A study of many of the Roman
inscriptions also shows a slight variation in
the treatment of the letters. Sometimes
the stroke moves above and below the cap
height, and sometimes one letter is placed
within another in an attempt to fit the
text into the available space on the stone.
The typeface Jupiter created by Patrick Griffin
is inspired by these inscriptions and includes a
number of alternative versions for each letter.
Figure 9.1. Universal alphabet. The
Bauhaus teacher Herbert Bayer strongly
argued for the elimination of the uppercase
alphabet, as demonstrated in part by his
geometric Universal alphabet.
Figure 9.2. The Trajan column. Completed
in AD , the lettering at the base of the
column is often viewed as one of the great-
est representations of the Roman capitals.
The German Bauhaus School announced in the early th century that for
maximum legibility, text should be set only in lowercase letters. As we know,
they did not succeed in their project, and the two writing hands are still with
us as separate sub-categories within the Latin alphabet. While the lowercase
alphabet is the workhorse carrying the biggest load, the uppercase alphabet
plays the important role of emphasising names and the beginning of new sen-
tences and of helping us navigate through the text. The two alphabets also
have very different foundations. The lowercase letters originate from a script
developed for continuous writing, while the uppercase letters derive from a
much slower, meticulous writing style handed down from Roman inscriptions.
This important difference in their origin has resulted in the two alphabets dif-
fering on several levels, both in their structure and in their skeletons.
Figure 9.4. Roman inscriptional proportions
in the typeface DTL Haarlemmer. This type-
face by Frank E. Blokland is a digitisation of
Jan van Krimpen’s serif type of . Blokland
has further added a sans serif version to the
family. The Roman inscriptional proportions
are to be found in many of the uppercase
alphabets by Jan van Krimpen.
Figure 9.7. Roman inscriptional proportions
in the typeface Garda. Inspired by the work
of the Renaissance writing master Francesco
Cresci, designer Mário Feliciano developed
Version One as a close interpretation of the
original. Versions Two and Three have the same
skeleton but different strokes.
Figure 9.6. Character proportions of
letters from early Roman inscriptions.
Wide The wide character group more or
less fits into a square. This group contains
the round letters ‘O’, ‘C’, ‘G’, ‘Q’, ‘D’ and
the letters with two vertical stems such
as the ‘N’, the ‘M’ (sometimes slightly
wider) and to some extent also the ‘H’
(often somewhat narrower).
Medium The medium-width character
group fits into a rectangle slightly nar-
rower than a square and contains letters
with diagonal strokes such as ‘A’ and ‘V’.
Narrow The narrow-character group fits
into a rectangle that is sometimes half as
wide as a square, and sometimes slightly
wider than half. Most of these letters are
divided into an upper and a lower part. To
maintain a circular shape in letters such
as ‘S’, ‘B’, ‘P’ and ‘R’, and thus correspond
to the circular shapes of the wide, round
letters, these letters will automatically
become narrow. The width of ‘B’ then
naturally defines the width of ‘E’, which
in turn defines the width of ‘F’, ‘L’ and ‘T’.
In that way, the goal of embedding the
circle in the letters automatically dictates
the width of the majority of the letters.
The proportions are illustrated with the type-
face Trajan Pro by Carol Twombly.
As mentioned, the basic shape of the uppercase letters emerged from the
Roman inscriptional capitals, with the Trajan column as the most refined
example. When the Renaissance humanists rediscovered the ancient cultures
of Rome and Greece, they developed a fascination with the Roman capi-
tals and a wish to define the correct proportions of letters based on these
inscriptions. As a result, many before Moxon had praised the idea of geometric
diagrams of circles and squares as constituting the ideal basic proportions. It
is said that when calligrapher and Renaissance scholar Felice Feliciano exam-
ined the ancient Roman inscriptions in , he could still see the marks
of the ruler and compass that the Romans had used to create the letters.
However, the humanist approach of basing all letters on the square and
the circle of the mathematically friendly -degree axis is not fully consist-
ent with the letters in the inscription on the Trajan column. In a way, these
Renaissance drawings appear to have been more of a reflection of the needs
of the humanists to explain structure in rational terms than a thorough study
to map out the proportions of the past. Although they followed the basic
grid of the Roman inscriptions, the narrow character group of the works of
the Renaissance printers was slightly wider than the Trajan model, and so the
uppercase letters became more uniform in structure than they had been in
ancient Rome.
Figure 9.5. Alphabet by Felice Feliciano.
Inconsistent with what is seen on the
Trajan column, where the round letters
are slightly oval, with an axis of about
degrees, Feliciano insisted in that
the axis of the ‘O’ should be degrees.
Figure .. The width of Univers. In the
comprehensive book from present-
ing his life achievements, Adrian Frutiger
argues that the differences in width of the
Roman inscriptional capitals is preferable
for achieving an optimal reading flow. He
nonetheless also points out that in a type-
face such as his own Univers, which features
multiple fonts, the proportions of the up-
percase letters must be evenly harmonised.
Figure 9.9. Modern interpretation of
th century jobbing types. The typeface
Aveny-T by Henrik Kubel, is inspired by the
large poster types of the th century.
Figure 9.8. Uniform Didone capitals. In the
Didone tradition, the uppercase letters have
approximately the same width. The typeface
Displaybaum by Matthew Carter is based on the
work of Justus Erich Walbaum (-).
Relative width
With the introduction of the Didone style, uppercase letters took on an
almost even width. Before this, the Transitional faces of Baskerville and
Fournier had taken a small step in this direction by slightly widening some of
the narrower characters. This fashion of even width eventually resulted in the
previous circular shapes no longer seeming evenly rounded throughout the
different letters; instead the uppercase characters developed into a mixture of
straight and curved lines.
Although the first known sans serif specimen printed in by William
Caslon IV shows traces of the Roman inscriptional proportions in the circular
‘O’, most early sans serif faces demonstrated a uniform optical width between
uppercase letters. This was a logical choice, since the Didone tradition defined
the norms at the time. Furthermore, the new jobbing types made for posters
needed to be printed in the largest point size possible. The square oval form
in the round letters made this possible by enabling words printed in uppercase
letters to appear as solid blocks of black and white, with an even texture. Yet
in the early th century, Edward Johnston, Paul Renner and other design-
ers reapplied the older Roman proportions in their sans serif designs, as they
argued that the uneven width enhances letter differentiation and improves on
word shapes. When asked whether the varying horizontal proportions would
not result in a lower readability than if the letters were of the same width,
Johnston’s reply was that an absolute dead evenness was a bad idea, and that
high readability is self-evident in these letters due to the characters’ ability to
demonstrate simplicity and distinctiveness at the same time.
Figure 9.11. Uppercase is superior to
lowercase. In a comparison of the same
point size, uppercase letters tend to perform
better than lowercase letters when read
from a distance and when blurred.
Illustrated with A2 Monday by Henrik Kubel.
Which is more legible?
It is a widely held belief that lowercase letters are more legible than uppercase
letters. This notion originates in the misconception that we read by identify-
ing word shapes; based on this assumption, a logical conclusion would be that
since uppercase letters have no ascenders or descenders, lowercase letters
become more distinctive and easier to recognise. Nonetheless, as discussed in
detail in Chapter , most perceptual psychologists today agree that we read by
a parallel process of registering the individual features of both the letters and
the words. If word shapes are less influential than commonly assumed, what,
then, is the internal legibility relation of uppercase and lowercase letters?
Figure 9.12. Uppercase similar to
lower case. In distance reading when the
upper case height is adjusted to match the
lowercase x-height, the two deliver similar
performances. Illustrated with A2 Monday by
Henrik Kubel.
Figure 9.13. The comparative legibility
level of uppercase and lowercase Univers.
For the uppercase and lowercase alphabets
to be equally legible in a search task, re-
search found that when the uppercase let-
ters has a height of . mm, the lowercase
letters should have a height of . mm.
. units
. units
A highly valued quality in continuous reading is the ability to see past the
printed letters and lose oneself in the content of the text. It is therefore no
surprise that many reading speed studies have found uppercase letters to
perform badly, as few readers are used to reading long passages of text set
in uppercase letters. Therefore, it will be difficult to get past the unusual
character of the text setting and get into the reading flow. Since we are not
used to reading all-caps text, reading longer paragraphs set in uppercase is
problematic. The issue is not quite as straightforward when it comes to read-
ing short words from afar. A review of the research on distance and search
reading suggests that uppercase is preferable in signage with limited vertical
space, while lowercase setting is preferable in signage with limited horizon-
tal space. Multiple studies of distance reading have compared uppercase and
lowercase letters in the same point size and found uppercase letters to be
more legible. The same goes for blurred text read at close range (Fig. .).
In a way, these results are not surprising, since uppercase letters are both
wider and taller than the x-height of their lowercase counterparts. However,
when a comparison is made between the two alphabets where the uppercase
height is adjusted to be the same as the lowercase x-height, the two deliver
similar performances (Fig. .). This still does not provide an ideal com-
parison, since the ascenders and descenders of the lowercase letters extend
beyond the x-height area, in that sense creating an advantage for lowercase
letters by taking up more vertical space. A fair legibility comparison between
the two alphabets is therefore no easy task. In attempts to define the ideal
internal size relationship for a search task method, research has found that
when uppercase Univers is . mm tall, the x-height of the corresponding
lowercase letters should be about . mm (Fig. .). With this size relation-
ship, the lowercase letters in most typefaces take up less horizontal space
and have the largest vertical size. Although the lowercase letters appear to
win the overall race between the two alphabets, all-uppercase text should not
be abandoned altogether. Some signage situations still benefit from the wide
letter shapes and the better use of the vertical area that can be achieved
with all-uppercase text.
Figure 9.14. Familiar and unfamiliar place
names. In a study of type for traffic signage,
researchers found no difference in perfor-
mance between all-uppercase words and
lowercase words with initial capitalisation,
when the words were previously unknown
to the participants. However, when testing
the recognition of already known words, the
initially capitalised lowercase words showed
better performance than all-uppercase
words. The researchers speculated that
this was due to the reader’s mental image
of place names, which are often lowercase
words with initial capitalisation; this makes
the cognitive task of matching easier for
initially capitalised lowercase words than for
all-uppercase words. The all-uppercase words
were set in the font Standard Highway Series D,
while the lowercase words were set in an early
version of the ClearviewHwy typeface, created
in .
1. Moxon, J. (1683) Mechanick Exercises or the Doctrine
of Handy-works, London: Joseph Moxon, vol.2, p.15.
2. Morison, S. (1927) ‘Introduction’, A Newly Discovered
Treatise on Classic Letter Design Printed at Parma by Dami-
anus Moyllus circa 1480, Paris: At the Sign of the Pegasus,
pp.9-23.
3. Johnston, E. (1986) Lessons in Formal Writing, H.
Child & H. Howes, (eds), Taplinger, p.92.
4. Frutiger, A. (2008) Adrian Frutiger Typefaces: The
Complete Works, H. Osterer & P. Stamm, Basel: Birkhauser.
5. Poulton, E.C & Brown, C.H (1968) ‘Comprehension
of an existing teleprinter input and of possible alterna-
tives’, Journal of Applied Psychology, vol.52(1), pp.16-21.
Tinker, M.A. & Paterson, D.G. (1928) ‘Influence of Type
Form on Speed of Reading’, Journal of Applied Psychology,
vol.12, pp.359-368.
6. Tinker, M.A. (1932) ‘The Influence of Form of Type
on the Perception of Words’, Journal of Applied Psychol-
ogy, vol.16(2), pp.167-74.
Arditi, A. & Cho J. (2007) ‘Letter case and text legibil-
ity in normal and low vision’, Vision Research, vol.47(19),
pp.2499-2505.
7. Burtt, H.E., & Bash, C. (1923) ‘Legibility of Bodoni,
Baskerville Roman, and Cheltenham type faces', Journal
of Applied Psychology, vol.7, pp.237-245
8. Smith, F., Lott, D. & Cronnell, B. (1969) ‘The effect
of type size and case alternation on word identification’,
American Journal of Psychology, vol.82(2), pp.248-253.
9. Poulton, E.C. (1972) ‘Size, style, and vertical spac-
ing in the legibility of small typefaces’, Journal of Applied
Psychology, vol.56(2), pp.156-161.
10. Garvey, M., Pietrucha, M.T. & Meeker, D. (1997)
‘Effects of Font and Capitalization on Legibility of Guide
Signs’, Transportation Research Record, National Academy
Press, Washington D.C., vol.1605, pp.73-79.
FHWA Series D
ClearviewOne CD ()
ClearviewHWY -W ()
References
10
Sans or
serif?
Arguments in favour of the serif
Type designer Gerard Unger has often spoken in favour of serifs. Unger argues
that due to the lack of details in the parafoveal view, the serifs on the
extremes of ascending and descending characters can help enhance the shape
of the words by adding emphasis to these parts of the letters (Fig. .). In an
article published on the Linotype website, Adrian Frutiger supports the notion
of serifs having the role of emphasising stroke endings; he further clarifies
that while a line without fortified endings will look incomplete and seem to
flow on forever, a line with fortified endings visually ends at its endpoint,
which helps define the letter shape (Fig. .). In , the scientist E.C.
Sanford explained the function of the serifs as protecting the ends of strokes
from the rounding effects of irradiation; he advocated the use of short serifs
since they can achieve this outcome alone, while serifs that are too long eas-
ily result in letter confusion.
In the Linotype article, Adrian Frutiger compares serifs to the top and bot-
tom terminations of columns in early architecture and argues that such ter-
minations have the function of keeping the columns optically apart and creat-
ing a rhythm between the shapes. This belief that serifs enable the reader to
identify the individual characters by keeping the letters apart is often voiced
among designers, reasoning that when any two characters of vertical strokes
like ‘i’ and ‘l’ are placed next to each other, serifs will keep them spaced out,
and so enhance perception of the individual characters.
A final argument in favour of the serif states that the horizontal strokes of
the serifs help the eye stay on the line of text and not wander off. The view
is supported by the principle of good continuation that was first defined by
the school of Gestalt psychologists in the early part of the th century. The
principle states that a number of objects placed in a row tend to be perceived
by the eye as connected elements flowing in a certain direction (Fig. .).
Based on this phenomenon, serifs could be claimed to help group letters into
words and thus enhance the horizontal reading direction.
Arguments such as these originate in logical reasoning and analysis of the
letter shapes. Nonetheless, if we look at the matter with a more scientific
approach, a very different picture emerges.
Figure 10.3. The end of the line. According
to Adrian Frutiger, any line needs fortified
endings to keep it from flowing on end-
lessly.
Statements such as “serifs are important for reading” or “always use a serif
typeface for setting longer paragraphs of text” have been repeated so often that
many simply consider them facts. Are they facts, or do serifs in fact weaken
the essential letter shape; would it appear more defined on its own? Many
arguments have been put forward in favour of the serifs, and many ideology-
driven statements have been proposed in support of the sans serif faces. The
matter is still up for debate, yet it is time to abandon any rigid belief that
serifs are always better and adopt a more nuanced view on the use of both
serif and sans serif typefaces.
Figure 10.1. Serifs improving visibility in the
parafoveal view. Gerard Unger argues that
serifs improve the fuzzy image of ascenders
and descenders in the parafoeveal view.
Figure 10.2. The Gestalt law of proximity.
The school of Gestalt psychology defined a
series of laws related to our visual percep-
tion. One of these suggests that a series
of elements in spatial proximity will be
perceived as being connected.
Scientific findings
A summary of existing data from experimental studies does not speak in
favour of either serif or sans serif typefaces. In most cases, the data shows
no significant difference, and when it does, there are just as many studies in
favour of the serif as in favour of the sans. In a thorough review of a large
number of studies on the subject, Ole Lund found in that nearly all
the research reviewed lacks internal validity. The problem in many of these
studies is that the typefaces used as test material are often quite different
in style, not only in the absence or presence of serifs but also in the strokes
and shapes of the characters. The differences between Old Style and Didone
serif typefaces or between Geometric and Humanist sans serif typefaces often
seem to be overlooked.
Since Ole Lund made his review, more recent research has been published
to address these problematic issues. In a study involving type designer Charles
Bigelow, the scientists used test material designed to minimise the disruptive
influence of other variables. The sans serif and serif fonts in the test mate-
rial originate in Bigelow’s typeface Lucida, with the only difference being that
one font has serifs while the other has none (Fig. .). The methodology of
the study was to have participants read rapidly presented words exposed on a
screen placed at a distance. The results demonstrated that participants found
it more difficult to read the slab serif font in small sizes than the sans serif
font; however, in larger sizes there was no difference between the two. Does
this mean that serif faces lower legibility in distance viewing? Perhaps, but
there might also be another explanation that relates only indirectly to the
serifs. To unify the spacing between the typefaces, the serif version had been
more tightly fitted than normal, while the sans serif version had been more
loosely fitted. Knowing this, it is possible that crowding of the serif letters
may have influenced the data (for more on crowding, see Chapter ).
Whatever the reason, the study tells us that serifs are not by default a
legibility-improving feature.
Figure 10.6. The serif and sans serif
versions of Lucida by Charles Bigelow.
An experimental investigation comparing
the two designs on screen found that the
serif version was more difficult to identify
in small sizes, while no difference between
the two was found in larger sizes.
Figure 11.5. Serifs as horizontal guidelines.
The serifs of the typeface FF Avance by Evert
Bloemsma are designed to guide the eye in a
forward direction, with solid shapes on the top
left of the stems and on the bottom right.
Figure 10.4. Serifs in webfonts. A matter
of importance to type designer David Berlow
of the Font Bureau is to reintroduce serif
typefaces for screen presentation, as he
finds that sans serif typefaces like Verdana
and Lucida Sans have wrongly been taking
over the web for a lot of text use. According
to Berlow, sans serif faces often represent a
‘simplified’ roman script, while serif faces, as
shown here on PoynterSerif RE, have a more
natural differentiation between characters,
which he finds suitable for small screen text.
Familiarity of the sans serif
Before the sans serif became a part of everyday life, traditionalist designers
and writers strongly opposed its application in reading material convention-
ally dominated by serif typefaces. Stanley Morison (-) fought against
this trend and growing threat as he perceived it, proclaiming that “the serif
is a device which centuries of experience of reading has made into a convention so
strong that the efforts of artist, intellectuals and educationalists have never been
strong enough to destroy it”.
It is remarkable that, even at a time when the sans serif was rather unpop-
ular amongst traditionalist designers, it performed quite well in test situa-
tions. A legibility investigation in found that the only sans serif in the
study ranked second, being % less legible than an Old Style face and hav-
ing % superiority over the third on the list, a Didone. Another study carried
out in compared different typefaces, including Kabel Light as the
only sans serif, and found that the difference between the sans serif and the
Old Style and Didone faces was not statistically reliable, and so the research-
ers concluded that the Kabel Light, Old Style, and Didone faces were equally
legible. Yet, when readers were asked to rank the same types based on their
preferences, Kabel Light was judged as the second most illegible of all .
Figure 10.8. Testing of the sans serif
Kabel Light. The test material used in
the many studies by Miles A. Tinker in-
cluded only one sans serif font, this be-
ing the geometric Kabel Light. In a study
of , the performance of Kabel Light
was equal to the Old Style and Didone
faces used in that instance.
Figure 10.9. The British traditionalists.
In the years of the Bauhaus school and the
New Typography movement spreading their
ideas in the German-speaking countries,
Stanley Morison and his fellow traditional-
ists in England were busy reviving a number
of ancient typefaces. Illustrated here with the
Francesco Griffo revival Bembo (c. ).
In , John Harris carried out a study comparing typefaces of different
styles. In a short exposure study presenting single letters to the parafoveal
view, Harris compared the legibility of the three typefaces Baskerville ,
Univers Medium and Gill Sans Medium, the first of these being a serif typeface
and the two latter being sans serif typefaces. Where most other comparative
studies look at the typefaces as a whole, Harris investigated the legibility of
the individual letter, and so came up with some interesting findings related
to the serifs of Baskerville. His results showed that the positive or negative
effects of serifs differ depending on the letter under investigation. If vertical
letters with multiple stems such as ‘h’, ‘n’, and ‘u’ have serifs on the inside
counters, the confusion between ‘b’ and ‘h’, and between ‘n’, ‘u’, ‘o’ and ‘a’,
are significantly more likely in Baskerville than in the two sans serifs. However,
in vertical letters of one stem, such as ‘i’ and ‘j’, the top serifs appeared to
enhance the differentiation from other characters by emphasising the gap
between the stem and the dot.
Figure 10.7. Serifs playing different roles.
Researcher J. Harris has shown that serifs
have different effects on legibility depend-
ing on the letter in question. Harris found
that when placed on the inside counters of
‘h’, serifs reduce the differentiation from ‘b’,
and when placed on the inside counter of ‘n’
and ‘u’, serifs reduce the differentiation from
other x-height letters such as ‘a’ and ‘o’. Yet
it appears that the top serif on ‘i’ enhances
the letter’s differentiation from ‘l’.
Illustrated with ITC New Baskerville.
So how did the sans serif typefaces manage to gain ground if readers did not
enjoy reading them? To understand this, it is relevant to take a look at the
ideology of the modernist movement that was spreading over Europe in the
early th century. The typographical foundation of this belief was expressed
by Jan Tschichold in his book The New Typography of , where Tschichold
stated that “it must be laid down that sanserif is absolutely and always better”.
He further argued that the sans serif types were the right type for the time,
as they symbolised the tendencies seen in modern architecture. Contemporary
English traditionalists strongly disapproved of the sans serif for continu-
ous reading and argued that the main role for the style was to be applied in
timetables and signage, not in running text. Tschichold himself was also not
too keen on sans serifs for continuous reading, admitting that he found serif
faces to be more legible for the purpose. Nonetheless, sans serif typefaces
eventually became the typographical symbol of the Modernist spirit, broadly
applied in all kinds of reading material.
This happened not in response to readers’ preferences but because of
designers’ conceptual beliefs; readers were exposed to the sans serif to such
a degree that they eventually got used to it.
Figure 10.10. Jan Tschichold. The revival
of Tschichold´s headline typeface ‘new
constructible block-script’ (ca. )
developed by Christopher Burke.
1. Frutiger, A. (2010) The History of Linear, Sans Serif
Typefaces [online], available from:
www.linotype.com/2258/introduction.html
2. Sanford, E.C. (1888) ‘The relative legibility of the
small letters’, American Journal of Psychology, vol.1(3),
pp.402-435.
3. Lund, O. (1999) Knowledge Construction in Typogra-
phy: The case of legibility research and the legibility of sans
serif typefaces, Thesis (PhD), The University of Reading,
Department of Typography & Graphic Communication.
4. Morris, R.A., Aquilante, K., Yager, D. & Bigelow, C.
(2002) ‘P-13: Serifs Slow RSVP Reading at Very Small
Sizes, but Don’t Matter at Larger Size’, SID Symposium
Digest of Technical Papers, vol.33(1), pp.244-247.
5. Harris, J. (1973) ‘Confusions in letter recognition’,
Professional Printer, vol.17(2), pp.29-34.
6. Morison, S. (1959) Introduction to Cyril Burt’s: A
psychological study of typography, London: Cambridge
University Press, p.xi.
7. Pyke, R.L. (1926) Report on the Legibility of Print,
Medical Research Council, London: His Majesty’s Station-
ery Office.
8. Tinker, M.A. & Paterson, D.G. (1932) ‘Studies of
Typographical Factors Influencing Speed of Reading, X.
style of type face’, Journal of Applied Psychology, vol.16,
pp.605-613.
9. Tinker, M.A. (1964) Legibility of Print, U.S.A.: Iowa
State University Press.
10. Tschichold, J. (1995) The New Typography, first
published in 1928, University of California Press, p.74.
11. Morison, S. (1999) Tally of Types, Boston: David R.
Godine, see p.98-99.
Monotype Recorder (1933) ‘An Account of The L·N·E·R
Type Standardization’, The Monotype Recorder, vol.32(4),
pp.7-11.
12. Kinross. R. (2004) Modern typography: an essay in
critical history, London: Hyphen Press, p.106.
References
11
Hanging
and ranging
numerals
Up until the Industrial Revolution, printed type was mostly text type. The
ranging numerals with digits of cap height first came into play with the
coming of the big, bold display faces of uppercase letters designed to yell
out from the page and clamour for the attention of passers-by. The hang-
ing numerals of varying height were problematic under these circumstances,
and so the ranging numerals became a popular choice for typesetters and
sign writers. These solid blocks of digits eventually became so popular that
they also became the standard in text type. As seen later in the specimen
books published by the American Type Founders at the beginning of the th
century, ranging numerals had become the norm to such a degree that the
expected hanging numerals of several revivals of the works of Caslon and
Baskerville were replaced with ranging numerals. Another solution is shown in
the Garamond Series (Fig. .); here the revival is furnished with semi-hanging
numerals of a size slightly below cap height. The normally x-height digits ‘’,
‘’ and ‘’ are now full size, and the ascending and descending digits extend
only slightly outside the main height. In this way, the typeface accommodates
the vogue at the time with evenly sized numerals, while respecting the Old
Style numerals of the original source.
Figure 11.3. The Garamond Series.
Published by the American Type Founders in
the specimen book of , this Garamond
revival is furnished with semi-hanging
numerals instead of the more historically
correct hanging numerals.
Figure 11.2. Spacing the ‘1’. To line the
numerals evenly on top of each other in
columns, ranging numerals traditionally
have the same width. This makes it difficult
to design the digit ‘1’ to match the others
without having too much space around
it. To avoid this problem, the tabular ‘1’ is
often broader than the proportional ‘1’.
Illustrated with the typeface Apud by Dino
dos Santos.
The Hindu-Arabic numerals joined the Latin alphabet between the th and
th centuries. While the lowercase letters are based on a running construc-
tion with the downstroke of one letter optionally merging into the upstroke
of the following letter, this is not possible in the numerals, which must
function as disparate entities. The difference is further evident in the stroke
endings of the Old Style figures ‘’ and ‘’ being diagonal and pointy, unlike
any of the Old Style letters, and it is evident in the flags of ‘’ and ‘’, both
appearing to be drawn with the broad side of the pen in contrast to the nar-
row horizontal strokes of the Old Style letters. Nonetheless, over the years,
the numerals developed into the style we know today as hanging numerals (or
Old Style and non-lining), which approximately matches the roman lowercase
letters – so that the ‘’, ‘’, ‘’ are of x-height, ‘’ and ‘’ ascend, and the rest
of the numerals descend. The more familiar ranging (modern or lining) numer-
als of same height as the capitals came much later.
Figure 11.1. Hanging and ranging numerals.
The hanging numerals fit the lowercase let-
ters by being both descending and ascending.
Ranging numerals all ascend, like the capitals.
Illustrated with the typefaces Merlo (top) and
Stella (bottom), both by Mário Feliciano.
abcp 1234567890
ABC
abcp
ABC 1234567890
Merlo
Stella
Proportional Tabular
Ranging
Hanging
Figure 11.6. A selection of hanging
and ranging numerals.
Eames Century Modern from House Industries A2 Brew Type Display by Henrik Kubel
A2 Vogue Floral by Henrik Kubel Musee by Dino dos Santos
So are hanging numerals more legible than ranging numerals? Several typo-
graphical scholars of the early th century made a case in favour of the
hanging numerals. While discussing the setting of numbers in lowercase
text and large tables, the influential printer Theodore Low De Vinne argued
that, due to the different sizes, one hanging digit could rarely be mistaken
for another3; the argument was that hanging numerals are more legible than
ranging numerals due to the extra space surrounding the smaller x-height
digits in contrast to the uniform height of ranging numerals, which take up
all the available space.
Based on the same argument, Legros and Grant reach the opposite con-
clusion in their book Typographical Printing Surfaces of , where they state
that “the old-style numerals are irregular, and, owing to the smallness of some
sorts, their legibility is no greater than that of the modern figures. Moreover, the
fact that they comprise ascenders, descenders and small sorts make them unsuit-
able for most scientific works. Old-style founts are therefore frequently ordered
with modern figures”4.
Figure 11.4. Semi-hanging in the type-
face Miller. The typeface Miller by Matthew
Carter follows a tradition found in the
Scotch Roman faces from the Miller and
Wilson foundries of the early th century
(also seen in Garamond), where the
numerals were shorter than the uppercase
letters, with the ‘’ being slightly ascending,
and the ‘’ and ‘’ being slightly descending.
This way, Carter avoids the numbers stand-
ing out too prominently in text setting,
while still maintaining a certain uniformity.
Figure 11.5. The width of the numerals.
The golden rule amongst typographers has
long been that all numerals should fit the
width of an en-dash. This was the recom-
mendation of typefounder Pierre Simon
Fournier in 1 and printer John Smith in
2. Later writers, however, have estab-
lished these proportions to be slightly too
narrow in small sizes and low-quality print,
where the ink tends to fill out the coun-
ters. As a result, founders of the early th
century often supplied alternative wider
numerals for small point sizes. Illustrated
with Garamond Premier Pro by Robert Slimbach,
where all hanging numerals have the width of
the en-dash.
En-dash