ArticlePDF Available

Abstract and Figures

This paper introduces an arithmetic formula for the calculation of text line length (also referred to as line width) for roman alphabet from 1) the length of the alphabet in lowercase, 2) a value for the desired character density and 3) a mathematical constant. A short-range study with this formula has shown a margin of error of less than 5% in common serifed text typefaces. The potential application of this formula in both print and digital editorial products could be diverse, from the approximate calculation of pages in a book to the establishment of control parameters in responsive web pages. Moreover, this formula would allow designers to make decisions about formal aspects on reading devices based on principles of readability and reading experience.
Content may be subject to copyright.
112
Visible Language
50.1
113
Line Length
Peña
Abstract
This paper introduces an arithmetic formula for the calculation of text line
length (also referred to as line width) for roman alphabet from 1) the length
of the alphabet in lowercase, 2) a value for the desired character density
and 3) a mathematical constant. A short-range study with this formula has
shown a margin of error of less than 5% in common serifed text typefaces.
The potential application of this formula in both print and digital editorial
products could be diverse, from the approximate calculation of pages in a
book to the establishment of control parameters in responsive web pages.
Moreover, this formula would allow designers to make decisions about
formal aspects on reading devices based on principles of readability and
reading experience.
KEY WORDS:
Typography, Editorial Design, Metrics, Line length, Character density
Calculating Line Length:
an arithmetic approach
Ernesto Peña
114
Visible Language
50.1
115
Line Length
Peña
average of characters per word between English and Spanish in narrative
and non-narrative genres:
Considering these ndings, within English alone, the charac-
ter density of 12 words from a non-narrative piece would be higher than
14 words from a narrative piece on two equally long text lines. It is likely
that this and other circumstances have provoked CPL to be largely favored
as a line length unit. Several contemporary authors and researchers (e.g.
Bringhurst, 2004; De Buen, 2014; Dyson, 2004) have employed this measure
routinely enough to claim that if there is something close to a standard in
regards to line length, it might be CPL.
The use of the character density as a criterion for the line length
is not free of challenges. Unlike with what happens with continuous data
units where the line length can be easily induced with any text processor or
self-publishing software, the resources for calculating the character density
are limited: from all the scholars that have embraced CPL as a unit for deter-
mining line length, only few (i.e., Bringhurst, 2004; De Buen, 2014) have pro-
vided resources for the calculation thereof. Bringhurst (2004) recommends
staying within a range of 45 to 75 CPL “for a single-column page set in a
serifed text face in a text size”, with 66 characters including spaces, “widely
regarded as ideal” (26). To induce these metrics, Bringhurst has proposed
a “Table of Average Character Count per Line” (29) that uses the length of
the lowercase alphabet (LCA) in typographic points (1/12 inches) set in the
typeface and font size that would be used in the text.
Although the origin of Bringhurst’s criteria is unclear, Nanavati
and Bias (2005) report on a study conducted by Dyson and Kipping (1998) in
which lines of 55 characters were perceived by the participants as easier to
read on screen than lines of 25 and 100 characters. In this case, the number
computed of characters does not include spaces. Nevertheless, if De Buen’s
calculations on the average of characters per word in English are accurate,
the inclusion of spaces in Dyson and Kipping’s would result in lines of be-
tween 68 and 70 (55 + [52 ÷ 4.09/3.42 – 1] characters, falling relatively close
to Bringhurst’s suggestion.
For his part, De Buen (2014) argues that the length of the text
line should depend not only on the requirements of the layout of the text
Introduction
The understanding of what makes a text readable has been a concern
across several elds of study since —at least— the last century, although
the records of the use of the term readability started by the rst half of the
XIX century (Michel et al., 2011; “Readability, N., 2014). By the mid 1960s, the
term readability was applied to dierent conceptions, referring on the one
hand to the ease of comprehension of the text either due to the complexity
of the topic or the writing style and on the other to formal aspects such as
the layout of the information or the typeface (Klare, 1963). Rather than con-
sidering the properties of the font such as size of ascenders and descenders,
x-height, color and stroke weight, the design of the serif and other distinc-
tive features, which have been encompassed within the concept of legibility
(Gaultney, 2001), the line length might be one of the most —if not the
most— relevant factor of readability in a text.
Nanavati and Bias (2005) reported the existence of at least 100
years of research on line length and gave an overview of the results of such
research from the studies conducted by Tinker and Paterson since the late
1920s to the impact thereof in digital media applications (Miles A. Tinker,
1928; Paterson and Tinker, 1929; M. A. Tinker and Paterson, 1929; M. A. Tinker
and Paterson, 1931; Paterson and Tinker, 1940; Paterson and Tinker, 1943;
Paterson and Tinker, 1946; Miles A. Tinker, 1963a; Miles A. Tinker, 1963b;
Miles A. Tinker, 1966). The results are diverse not only in respect to the data,
but also respect to the format that researchers have historically employed to
express them, given either in continuous data units such as millimeters, cen-
timeters, or picas, or in discreet data units such as words per line (WPL) or
characters per line (CPL). The implications of the choice of units go beyond
the mere format. As a readability value, the metric length of the line itself is
trivial unless it is combined with others such as the width of the characters,
the font size (Bringhurst, 2004, 27), or the quantity of words or characters
that are contained within that length. If any, the information that could be
considered from this sole value would be the reading area and the necessity
of the reader to follow up with the head while reading. However, even the
latter would require other data such as the distance between the reader and
the reading space or the font size. In contrast, values such as the number
of words or characters, also known as word- or character density” (Dyson,
2004, 379), would be independent of the chosen typeface or its metrics.
Inherently more convenient than the continuous data units,
and despite the fact that they have been used interchangeably (e.g.,
Spencer, 1969, 35), there are crucial dierences between the use of WPL
and CPL. For instance, in the case of the former, the number of words that
a xed line length could contain would depend on the average of letters
per word, a value that would vary between languages and even genres. A
self-conducted study by De Buen (2014) reported the a dierence in the
English
Spanish
Narrative
3.46
3.92
Non-narrative
4.09
3.97
TABLE 1.
Difference in average letters
per word between English
and Spanish, in narrative
and non-narrative (De Buen,
2014, 157)
10 12 14 16 18 20 22 24…
90 36 43 50 57 64 72 79 86…
TABLE 2.
Fragment of the table of
average character count
per line by Bringhurst
(2004). The extreme left
column indicates the
approximate length of
the LCA in points (90).
The following numbers
of that row indicate the
approximate character
density (36, 43, 50…),
and the top row indicates
the approximate length
of the text line in picas
(10, 12, 14…). The
original table includes an
indication of the ranges
of CPL recommended by
Bringhurst.
116
Visible Language
50.1
117
Line Length
Peña
but also on dierent factors among which should be the prociency of the
reader, which would be gauged within a range between novice and expert1.
According to De Buen, the character density in a document for a novice
reader should range between 34 and 60 characters (including spaces) with
45 as optimal. A document for an expert reader should range between 45
and 80 characters with 60 as optimal. The method oered by De Buen (2014)
for the calculation of line length from the character density suggests to
obtain the value of the LCA and subsequently multiply it by 1.75 to obtain
the optimal line length, denominated l. This new value should be then multi-
plied by 0.75 to determine the minimum length, denominated n, and by 1.5
to determine the maximum, denominated m. Arguably, the origin of these
values respond to an approximate extrapolation from the 26 characters of
the LCA:
Optimal 26×1.75=45.5 45
Minimum 45.5×0.75=34.125 34
Maximum 45.5×1.5=68.25 60
I would argue that despite the eectiveness of these resources
for the calculation of the line length, they still posit a few operative limita-
tions. On the one hand, the “Table of Average Character Count per Line
(Bringhurst, 2004, 29) is comprehensive enough to cover a broad range of
possible scenarios: from 10 to 160 CPL, 80 to 360 points of LCA (in incre-
ments of 5 points) and 10 to 40 picas of line length (in increments of 2 picas).
However, the calculation of the line length depends completely on the ac-
cess to the table. On the other hand, the mathematical formulas introduced
by De Buen (2014, 216) give more freedom to the designer, but its reach
is limited to the prescribed ranges. The proposal presented here seeks to
provide one possible resource to overcome these limitations.
A Proposal
What I propose is an arithmetic formula that could be used to calculate the
metric space within which a particular character density can be applicable.
This formula considers dierent scenarios in print or digital media, regard-
less of the chosen criteria for the denition of the character density within a
text line. It is constituted by three components: the LCA of the chosen type-
face at the size and spacing that would be used including the space charac-
ter (LCA’), the desired character density (Cρ), and a mathematical constant
that I have provisionally denominated S. This formula could be expressed as
follows: LCA × Cρ(S) = Ll. These components are intended to take account of
the horizontal metric features of the chosen font (width, size), the linguistic
features of the text and the criterion of the designer. The denominated LCA
and the S constant are described below.
EXTENDED LOWER CASE ALPHABET
LC A’
As with the previously mentioned devices (Bringhurst’s table, De Buens
formulas), the formula presented here employs LCA as a starting point. The
idea of using the length of the lower case alphabet as a reference for the es-
timation of the eciency of a typeface is not new: Legge and Bigelow (2011)
have qualied it as “a traditional typographic measure” (4) which inclusion in
typographic specimens used to be a common practice (e.g., American Type
Founders, 1953, 1968; Mergenthaler Linotype Company, 1951) along with
—in some cases— the characters that a pica could contain. Traditionally,
the LCA is obtained by writing the basic characters of the Roman lowercase
alphabet without accents, diacritics or digraphs [abcdefghijklmnopqrstu-
vwxyz] (Bringhurst, 2004; Legge and Bigelow, 2011) and measuring the
length of the string of characters, whether with physical instruments or digi-
tally. I would assume that the aim of this exercise is to consider all the dier-
ent widths that the characters in the lower case alphabet have. Therefore, if
that is the case, even when there are languages in which the basic alphabet
includes digraphs or accented characters, when digraphs are combinations
of already existing characters, it would be assumed that they are metrically
identical to another one and omitted. Accented characters would be omit-
ted as well as accents; diacritics do not usually aect the width of the charac-
ter, and their inclusion in the LCA string would yield duplicated values2.
However, in cases in which the alphabet or idiomatic practices of a given
language include letters that are not part of the previously introduced string
of characters (e.g., æ, ß, or l·l), such characters would have to be included
within the LCA string for they have a width of their own. Based on these cri-
teria and on the fact that the space character might be the most common in
written manifestations of practically every language3, I would argue that the
LCA should include the space character as well. I refer here to this extended
version of the LCA as LCA. The inclusion of non-Roman and space characters
and the way their inclusion would aect the formula is discussed below.
THE S CONSTANT
The third component of the formula is a number that derives from a version
of the LCA that includes the space character (LCA’), and therefore, it would
change depending on the features of the basic alphabet of the language in
1 De Buen (2014) uses the labels bajo lector and alto lector, translated literally as ‘low reader’
and ‘high reader’, respectively (221).
2 For instance, in Spanish, the ch and ll used to be considered part of the alphabet until its
recent removal. Their inclusion in the LCA string would yield an incidence of three l (l and
ll), two c and two h (c, h and ch). In turn, the ñ is still considered a letter of the alphabet, but
being metrically identical to an n, its inclusion would duplicate a width metric in the string
of characters (“Exclusión de ‘ch’ y ‘ll’ del abecedario,” n.d.).
3 De Buen (2014) reports that this is surely the case in Spanish and English.
118
Visible Language
50.1
119
Line Length
Peña
which the text is laid out. However, this constant is intended to be calcu-
lated only once for each scenario and used recursively, that is, once for basic
Roman alphabet, once for German, etc. Here, for illustrative purposes, I
focus mainly on the basic Roman alphabet, but the same principle could be
applied to any other. The S constant is the result of counting the inter-word
and inter-character metric spaces (as opposed to the optical spaces) within
the LCA as characters themselves, and dividing the resulting number by 1.
The outcome of this operation is a number that multiplied by
the LCA’ would give the average width of a single character considering
the inter-character space, providing a unit accountable for both characters
and space by treating space as if it was a character and getting an average
of both categories (characters and space). The procedure followed for the
acquisition of this number is described as follows:
1. The LCA was composed with a given typeface and its length
measured.
2. The metric spaces between characters were removed to
obtain a new inter-character space-less LCA’ denominated LCA’’.
The average width of the characters in LCA’’ is determined and
subsequently used to measure the metric dierence between
LCA’’ and LCA’.
3. This dierence, determined in LCA’ average characters, is
added to the LCA’ characters.
4. This new number of characters divides one. The 26 characters
of the alphabet and the space sum 27 characters, but as the in-
ter-character space units depend on the design of the typeface,
it would vary between fonts. I have found that, among profes-
sional serif fonts for continuous reading, this number tends to
be close to two, which would give a total of 29 characters. This
value responds to a general estimation to which I have arrived
after applying the previously described protocol to several
typefaces, and it is by no means exact or infallible. During my
own tests I have found typefaces in which this number is closer
to one (e.g., MT Dante) or to three (e.g., Fedra serif A, Proforma),
but I still have not found any typeface that yields an average of
inter-character space smaller or larger that this range. However,
it is worth pointing out that despite the variance of this value
in some typefaces, the formula introduced here presents a rela-
tively small margin of error (±5%), even when applied to type-
faces in which the number resulting of the previously described
operation is closer to 1 or 3, including those mentioned before.
The best possible scenario might be to calculate this number
for every typeface used in a document and to keep a personal
record, but I would consider the formula provided here is —as it
is— a fairly good starting point.
The rounded quotient of the division of 1 by 29 results in
what I have denominated the S constant: 0.0345, which arguably applies
to most serif typefaces in languages with basic roman alphabet, in atten-
tion to the exceptions described in Extended Lower Case Alphabet (LCA’).
This number multiplied by the LCA and the desired character density (the
amount of characters that the designer wants to t in the text line) give the
length needed to t the required character density in the same units of the
LCA’. The number of characters that the line length result of the application
of the formula would t responds to the criteria of the designer and the
requirements of the text and its format, hopefully informed by the pertinent
research. For instance, to calculate the line length to t a desired density of
characters of Proforma at a size of 10 points would require laying out the 26
characters of the Roman alphabet plus one space character (LCA’)4. Assum-
ing that the LCA’ is 126.15 pt. (10p6.15) and that the desired density is 80
characters, the operation would be as follows:
126.15 pt. × 80(0.0345) = 348.17 pt. (29p0.17)
Applying a density of 40 to the same data, the result would be:
126.15 pt. × 40(0.0345) = 174.08 pt. (14p6.08)
For an extended Roman alphabet, or the use of characters
beyond the 26 of the Roman alphabet such as the German eszett (ß), a
similar criterion could be applied, but it would require to add the extra char-
acters to the LCA and to modify the S constant by adding the extra number
of characters. Applying these changes to the previous example, the LCA of
FIGURE 1.
On the left, the inter-
character metric space. On
the right, inter-character
optical space.
FIGURE 2.
First row: Lower case
alphabet plus the space
character, denominated
LCA’. Indicated in gray is
the inter-character spacing.
Second row: The calculation
of the average of the LCA’
without inter-character
spaces, denominated LCA’.
This average is subsequently
used to measure the inter-
character spacing.
LCA’
LC A’’ = y
4 For ease of measurement, the space character added to the LCA should be anywhere
within the string of characters except for the beginning or the end.
120
Visible Language
50.1
121
Line Length
Peña
132.10 (11p0.10) (abcdefghijklmnopqrstuvwxyz ß) laid out in Proforma 10,
the S constant would have to be modied to a rounded 0.0333 to consider
the new character (1/30). Assuming that the desired densities are 80 and 40,
the operations would be as follows:
132.10 pt. × 80(0.0333) = 351.91 pt. (29p3.91)
132.10 pt. × 40(0.0333) = 175.95 pt. (14p7.95)
A Short-range Study
To test the eectiveness of the formula presented here, I performed a simple
study: I took the rst 20 serif typefaces for continuous reading that I found
installed in my computer at the moment (most of them system fonts) and
applied the formula to calculate the line length for a density of 65 and 45
characters, values chosen for mere illustrative purposes. The protocol was
applied to a single paragraph of a placeholder text consisting of 6500 char-
acters obtained from a public website (“Lorem Ipsum”). In every case, the
text was laid out ragged (aligned to the left) at a size of 10 pt., and the aver-
age was obtained by dividing the number of characters by the number of
lines. As the purpose was to measure the density of characters within a line,
the last line was omitted from the average calculation when the text did not
fully reach the right margin. The results of this exercise showed that all the
operations stayed within a margin of dierence of 4% between the desired
density and the character density resulting of the calculation.
Conclusion
The formula presented here does not pretend to be the ultimate resource
to line length calculation from character density; I would consider it to be a
reference instead. There are in fact conditions for which this formula or any
other resource might not be eective, such as justied text or very low char-
acter densities. Regardless of whether or not this formula is helpful or accu-
rate in the conditions for which it was meant, the factors that might have an
inuence on the character density, and therefore on the line length, might
be too many to take into account in a single arithmetical resource. A few of
these factors have been already listed here, and I would argue that some
of them are circumstantially taken into consideration within this formula,
but there are many others that would require more complex protocols. A
particularly complicated factor that has been only partially discussed here
TABLE 3.
Results of applying the
formula to 20 serif fonts for
continuous reading: The first
column lists the typeface,
the second lists the value
of the LCA’, the third and
fourth, the percentage of
deviation from the desired
character densities, 65 and
45 respectively.
FIGURE 3.
A scatterplot based on the
results of the study. On the
x axis, the deviation of the
result of the application of
the formula for a desired
density of 45 characters
per line. On the y axis, for
a desired density of 65
characters. The inner square
delimits a 5% of deviation.
As shown in this table,
most of the results of the
operations fell into slightly
higher numbers, but never
over 5%.
Typeface
LCA' pt.
(%) 65 C
(%) 45 C
Adobe Caslon
119.60
1.84
0.84
Adobe Garamond
115.70
1.74
-0.07
Baskerville
119.60
1.83
0.46
Cambria
128.60
2.96
1.19
Century
137.80
-0.26
0.33
Chaparral
122.80
1.84
1.19
Charter
129.90
0.73
-0.22
Dante MT
118.08
3.01
2.39
Fedra Serif A
151.70
-0.03
-0.22
Fournier
110.70
3.29
1.66
Georgia
132.00
0.73
0.27
Hoe er Text
125.80
1.35
0.49
ITC Mendoza
130.60
-0.67
-1.47
Mercury G1 roman
130.60
3.29
2.43
Minion
120.90
2.97
1.19
Palatino
135.40
3.29
2.58
Proforma
126.90
3.97
2.58
PT Serif
132.90
0.73
-0.07
Scala
125.45
0.73
-0.90
Times New Roman
121.75
2.96
1.32
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
-7
-6
-4
-3
-2
-1
0
1
2
3
4
5
6
7
122
Visible Language
50.1
123
Line Length
Peña
is language. Bringhurst (2004) has addressed the fact that the features of a
particular language have (or should have) an impact on the way a text is laid
out. According to Bringhurst (2004), highly inected languages like Spanish
would require less inter-word space than less inected languages such as
English or German. This feature would be easily taken into consideration in
the formula presented here as the space is part of the string of characters
that compose the LCA. De Buen (2014) has pointed out that because the
frequency of use of characters varies among languages, this would have an
impact in the calculation of the character density. To be able to accommo-
date this factor within the formula presented here, the LCA’ would have to
consider not only the width of the characters but also their frequency, which
might be doable by measuring the characters widths individually and modi-
fying this number accordingly to frequency. However, this would result in a
very intricate method. The dierence between such hypothetical method
and the one presented here might be minimal, although this is subject of
further investigation. I would argue that beyond the formula itself, having
a dynamic arithmetic approach might well open a door to other possibili-
ties, such as the development of digital tools which could easily take into
account factors that seem too problematic to be considered for a shorthand
method, as this formula pretends to be.
Future endeavors
De Buen’s (2014) contentions on the relation between the reader experi-
ence and the character density within a text line is appealing and a possible
avenue for further research, but until this research is given and its results
published, it might be possible to focus not on the reader but on the
intended reader by gauging the readability level of specic content through
readability formulas. In a recent study, Begeny and Greene (2014) tested 8
of the most popular readability formulas for determining their eectiveness
in calculating the diculty of reading materials. The ndings show that de-
spite extensive use in several elds, the success of their sample of formulas
on such a task is questionable except for the Dale-Chall formula: Grade =
(0.1579 × percent unfamiliar words) + (0.0496 × word/sentence) + 3.6365.
The only resource outside of the text itself that this formula employs is a list
of 3000 words publicly accessible (e.g., “Dale-Chall Easy Word List Text File”
2014); there are online resources for the calculation of readability by this for-
mula (e.g., Scott, n.d.) and others. According to Begeny and Greene (2014),
this formula was identied as “a valid measure of text diculty level” (210)
from grade 4 and above. I would argue that whatever the formula employed
for its calculation by nding the relation between readability and character
density and applying this criterion to the desired character density value
(Cρ) of the formula introduced here, the two understandings of readability
presented at the beginning of this paper (namely ease of comprehension
and formal aspects) could potentially converge in a single device such as an
digital application that could calculate not only the readability of a text but
also the line-length that would be appropriate for laying out such content.
References
1923 American Type Founders Specimen Book & Catalogue. (1923). Retrieved
from http://archive.org/details/1923AmericanTypeFoundersSpec
imenBookCatalogue
American Type Founders. (1953). Spartan Family. digital reprint by www.
CircuitousRoot.com. Retrieved from http://archive.org/details/
ATFSectionCSpartanFamily
American Type Founders. (1968). Foundry Univers. digital reprint by www.
CircuitousRoot.com. Retrieved from http://archive.org/details/
ATFFoundryUniversDebernyPeignotIP1968
Begeny, J. C., & Greene, D. J. (2014). Can Readability Formulas Be Used to Suc-
cessfully Gauge Diculty of Reading Materials? Psychology in the
Schools, 51(2), 198–215. http://doi.org/10.1002/pits.21740
Bringhurst, R. (2013). The elements of typographic style (4th ed.). Vancouver,
B.C. Canada: Hartley & Marks, Publishers.
Burke, V., & Greenberg, D. (2010). Determining Readability: How to Select
and Apply Easy-to-Use Readability Formulas to Assess the Dif-
culty of Adult Literacy Materials. Adult Basic Education & Literacy
Journal, 4(1), 34–42.
Burt, C., Cooper, W. F., & Martin, J. L. (1955). A Psychological Study of Typogra-
phy. British Journal of Statistical Psychology, 8(1), 29–56. http://doi.
org/10.1111/j.2044-8317.1955.tb00160.x
Dale-Chall Easy Word List Text File. (n.d.). Retrieved from http://countword-
sworth.com/blog/dale-chall-easy-word-list-text-le/
De Buen, J. (2014). Manual de diseño editorial (4th ed.). Gijón, España: Edicio-
nes Trea, S.L.
Dyson, M. C. (2004). How physical text layout aects reading from screen.
Behaviour & Information Technology, 23(6), 377–393. http://doi.or
g/10.1080/01449290410001715714
Dyson, M. C., & Kipping, G. J. (1998). Exploring the eect of layout on reading
from screen. In R. D. Hersch, J. André, & H. Brown (Eds.), Electronic
Publishing, Artistic Imaging, and Digital Typography (pp. 294–304).
124
Visible Language
50.1
125
Line Length
Peña
Springer Berlin Heidelberg. Retrieved from http://link.springer.
com/chapter/10.1007/BFb0053278
Exclusión de “ch y “ll” del abecedario. (n.d.). Retrieved September 4, 2014,
from http://www.rae.es/consultas/exclusion-de-ch-y-ll-del-
abecedario
Gaultney, V. (2001). Balancing typeface legibility and economy. Retrieved
from http://www-01.sil.org/~gaultney/BalanLegEcon.pdf
Klare, G. R. (1963). The measurement of readability. Ames : Retrieved from
http://hdl.handle.net/2027/mdp.35128001330503
Legge, G. E., & Bigelow, C. A. (2011). Does print size matter for reading? A
review of ndings from vision science and typography. Journal of
Vision, 11(5). http://doi.org/10.1167/11.5.8
Lorem Ipsum. (n.d.). Retrieved from www.lipsum.org
Mergenthaler Linotype Company. (1951). Corona [typeface]. digital reprint
by www.CircuitousRoot.com. Retrieved from http://archive.org/
details/LinotypeCorona1951try2
Michel, J.-B., Shen, Y. K., Aiden, A. P., Veres, A., Gray, M. K., Pickett, J. P., …
Aiden, E. L. (2011). Quantitative Analysis of Culture Using Mil-
lions of Digitized Books. Science, 331(6014), 176–182. http://doi.
org/10.1126/science.1199644
Nanavati, A. A., & Bias, R. G. (2005). Optimal Line Length in Reading—a
Literature Review. Visible Language, 39(2), 121–145.
Paterson, D. G., & Tinker, M. A. (1929). Studies of typographical factors
inuencing speed of reading. II. Size of type. Journal of Applied
Psychology, 13(2), 120–130. http://doi.org/10.1037/h0074167
Paterson, D. G., & Tinker, M. A. (1940). Inuence of line width on eye move-
ments. Journal of Experimental Psychology, 27(5), 572–577. http://
doi.org/10.1037/h0054498
Paterson, D. G., & Tinker, M. A. (1943). Eye movements in reading type sizes
in optimal line widths. Journal of Educational Psychology, 34(9),
547–551. http://doi.org/10.1037/h0063086
Paterson, D. G., & Tinker, M. A. (1946). The relative readability of newsprint
and book print. Journal of Applied Psychology, 30(5), 454–459.
http://doi.org/10.1037/h0054012
readability, n. (n.d.). OED Online. Oxford University Press. Retrieved from
http://www.oed.com/view/Entry/158852
Sassoon, R. (2002). Computers and Typography 2. Intellect Books.
Scott, B. (n.d.). The New Dale-Chall Readability Formula. Retrieved Septem-
ber 25, 2014, from http://www.readabilityformulas.com/new-
dale-chall-readability-formula.php
Spencer, H. (1969). The visible word. (Royal College of Art (Great Britain), Ed.)
(2nd ed., revised). London: Lund Humphries in association with
the Royal College of Art.
Tinker, M. A. (1928). The Relative Legibility of the Letters, the Digits, and of
Certain Mathematical Signs. The Journal of General Psychology,
1(3-4), 472–496. http://doi.org/10.1080/00221309.1928.9918022
Tinker, M. A. (1963a). Inuence of simultaneous variation in size of type,
width of line, and leading for newspaper type. Journal of Applied
Psychology, 47 (6), 380–382. http://doi.org/10.1037/h0043573
Tinker, M. A. (1963b). Legibility of print. Ames: Iowa State University Press.
Tinker, M. A. (1966). Experimental Studies on the Legibility of Print: An An-
notated Bibliography. Reading Research Quarterly, 1 (4), 67–118.
http://doi.org/10.2307/747222
Tinker, M. A., & Paterson, D. G. (1929). Studies of typographical factors
inuencing speed of reading. III. Length of line. Journal of Applied
Psychology, 13 (3), 205–219. http://doi.org/10.1037/h0073597
Tinker, M. A., & Paterson, D. G. (1931). Studies of typographical factors inu-
encing speed of reading. V. Simultaneous variation of type size
and line length. Journal of Applied Psychology, 15(1), 72–78. http://
doi.org/10.1037/h0073704
Author
Ernesto Peña is an editorial and information designer, design researcher
and educator. His current research interests include knowledge mobiliza-
tion, digital humanities, and visual literacy with interest in experimental
design and typography since the beginning of his professional exercise as a
designer in 2000. Ernesto has worked as an educator in Mexico and Canada,
and as a professional designer in Mexico and Germany. Ernesto is currently
a researcher for the Implementing New Knowledge Environments project
(INKE), involved mainly in text visualization interface design, and a research
associate for the Digital Literacy Centre in the Faculty of Education at the
University of British Columbia.
PhD Candidate
Department of Language and Literacy Education,
Digital Literacy Centre
Faculty of Education
University of British Columbia
ResearchGate has not been able to resolve any citations for this publication.
Article
Résumé La lisibilité relative des lettres, des chiffres, et de certains signes mathématiques a été déterminée au moyen d'une technique de courte exposition où le pourcentage des lectures correctes a donné le résultat de lisibilité. La Série I a compris les lettres minuscules plus les autres caractères, et la Série II a compris les majuscules plus les chiffres et les signes. La lisibilité des lettres comme déterminée dans cette étude et dans les études par d'autres a été comparée avec la fréquence des lettres dans l'impression. L'ordre de la lisibilté des lettres a été à peu près le même conniaté dans les études antérieures (sauf quand on a employé des types de test). L'ordre de lisibilité des chiffres est un peu différent dans les deux séries mais il n'y a qu'une petite différence au cas des signes mathématiques. La plupart des chiffres tombent dans le groupe des lettres ayant une assez bonne lisibilité et à peu près tous les signes sont groupés avec les lettres d'une pauvre ou assez bonne lisibilité. L'analyse montre que les facteurs qui influencent la lisibilité des caractères isolés sont: (a) grandeur, (b) simplicité ou complexité de contour, (c) épaisseur du bâton et du front du caractère, (d) les ombres et les lignes minces, (e) espace blanc renfermé entre les contours, (f) force ou manque de force dans les parties faisant des distinctions. Puisque l'usage, l'économie de l'espace dans l'impression, le placement de l'argent, et le goût esthétique empêchent des changements dans l'impression, il vaut mieux sans doute essayer, en cherchant des améliorations, de simplifier les contours, de diminuer le nombre de lignes minces, et d'appuyer sur les parties caractéristiques plutôt que de soutenir des changements radicaux de grandeur et de forme. La fréquence se trouve avoir une moyenne corrélation négative ou nulle corrélation avec la lisibilité. Quelques-unes des lettres employées le plus fréquemment sont d'une lisibilité très médiocre. Referat Die verhältnismässige Lesbarkeit der Buchstaben, Ziffern, und gewisser mathematischer Zeichen wurde durch eine kurze, mechanische Exponirung bestimmt, wobei der Prozentsatz der Häufigkeit des Richtiglesens, den Grad der Lesbarkeit ergab. Serie I umfasste die kleinen Buchstaben nebst allen andern Zeichen, und Serie II bestand aus den grossen Buchstaben sammt den Ziffern und Zeichen. Den Grad der Lesbarkeit der Buchstaben, wie man ihn nach dieser Untersuchung, sowie nach den Untersuchungen Anderer, festgesellt hatte, verglich man mit der Häufigkeit des Vorkommens derselben im Druck. Der Grad der Lesbarkeit der verschiednen Buchstaben stimmte so ziemlich mit den Ergebnissen früherer Untersuchungen überein, (ausser in Fällen wo man Test Typen benutzt hatte). Der Grad der Lesbarkeit der Ziffern varriiert etwas in den zwei Serien; bei den mathematischen Zeichen, jedoch, ist dieser Unterschied sehr gering. Die meisten Ziffern fallen in die Gruppe der Buchstaben von mittelmässiger Lesbarkeit und fast alle Zeichen kommen zu der Buchstabengruppe von mittelmässiger oder niedriger Lesbarkeit. Die Analyse beweist dass die Lesbarkeit einzelner Zeichen durch folgende Faktoren beeinflusst wird: (a) Grösse, (b) Einfachheit oder Verwicklung der Umrisse, (c) Breite der Striche und der Druckform, (d) Schattirung und Haarstriche, (e) der weisse Flächeninhalt der Umrisse, (f) Betonung der Unterschiede, oder Mangel an Betonung der Unterschiede der verschiedonen Teile. Da der herkömmliche Gebrauch, Oekonomie der Druekfläche, Unkosten und Schönheitssinn, Veränderungen des Druckes verhindern, so ist es wohl am besten, Veränderungsversuche auf die Vereinfachung der Linien zu beschränken, wie Verminderung der Haarstriche, und Betonung der charakteristischen Teile, als radikale Veränderungen in Grösse und Form zu befürworten. Häufigkeit des Vorkommens hat nur eine mässige, negative Korrelation oder gar keine Korrelation mit der Lesbarkeit. Manche Buchstaben die am häufigsten vorkommen, sind von sehr geringer Lesbarkeit.
Article
A grade level of reading material is commonly estimated using one or more readability formulas, which purport to measure text difficulty based on specified text characteristics. However, there is limited direction for teachers and publishers regarding which readability formulas (if any) are appropriate indicators of actual text difficulty. Because oral reading fluency (ORF) is considered one primary indicator of an elementary aged student's overall reading ability, the purpose of this study was to assess the link between leveled reading passages and students’ actual ORF rates. ORF rates of 360 elementary-aged students were used to determine whether reading passages at varying grade levels are, as would be predicted by readability levels, more or less difficult for students to read. Results showed that a small number of readability formulas were fairly good indicators of text, but this was only true at particular grade levels. Additionally, most of the readability formulas were more accurate for higher ability readers. One implication of the findings suggests that teachers should be cautious when making instructional decisions based on purported “grade-leveled” text, and educational researchers and practitioners should strive to assess difficulty of text materials beyond simply using a readability formula.
Article
There are many readability tools that instructors can use to help adult learners select reading materials. We describe and compare different types of readability tools: formulas calculated by hand, tools found on the Web, tools embedded in a word processing program, and readability tools found in a commercial software program. Practitioners do not need expensive software or extensive training to determine reading levels of materials. We include instructions for using the Flesch Reading Ease, Flesch-Kincaid, SMOG, Dale-Chall, Spache, FORCAST, Fry, and RIX readability formulas. (Contains 4 tables.)
Article
One of the most important, and most studied, aspects of human perception is the act of reading. Reading has received much attention from researchers, both from a human information processing (HIP) approach and as a common, practical act that needs to be optimized, especially in the realm of human-computer interaction (HCI). One of the text variables that has been studied for over 100 years is line length, at times referred as line width. Psychologists, typographers and others working in the field of reading and advertising have demonstrated the effects of line length on readability of text. Two of the questions addressed in past studies include, How long should a column of text be, to optimize readability of the text? and Which view is more preferred by readers--multiple narrow columns or one wide column with the same amount of information content? Research has led to recommendations that line length should not exceed about 70 characters per line. The reason behind this finding is that both very short and very long lines slow down reading by interrupting the normal pattern of eye movements and movements throughout the text. In a world of personal digital devices (PDAs), one-inch cell phone displays and of wide-screen TVs and full-wall computer displays, the question of line length has renewed timeliness. Studies reviewed here show that different aspects of reading performance such as comprehension, reading speed, method of movement (e.g., paging and scrolling) and eye movements are affected by changes in line length. In addition to that, various typographic factors such as font type and size, line and character spacing as well as different screen structures such as varying number of columns and screen sizes also affect readability. These factors have an effect on optimal line length for the text read from printed or on-screen material.
Article
The writers continue previous work, using the Chapman-Cook Speed of Reading tests. Speed of reading records obtained from 320 college students for texts set up in 6-, 8-, 10-, 12-, and 14-point type with line length constant at 80 mm. show that 10-point type yields the fastest reading. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Article
Five groups of 80 students each read selections from the Chapman-Cook Speed of Reading Tests set in type varying in size (6-14 pt.) and length of line (16-27 picas, respectively). The smallest text (6 pt. on 16 pica line) was read significantly more slowly than the standard (10 pt. on 19 pica); the next larger (8 pt. on 17 pica) was approximately equivalent to the standard; the two largest (12 pt. on 23 pica, and 14 pt. on 27 pica) were read slightly more slowly (difference not statistically reliable) than the standard. Neither size of type nor length of line can be considered as an independent variable in determining optimal set-up. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Article
In a series of 11 reading performance experiments it was demonstrated that excessively short lines (9 picas) and excessively long lines (43 picas) are both read much more slowly than lines of moderate width (19 picas). Eye-movement photographs of subjects indicate the precise changes in oculomotor patterns involved in reading excessively short and long line widths. It appears that the decreased efficiency with which a very short line is read is due to the reader's inability to make maximum use of horizontal peripheral cues. Analytical records derived from the reading of extremely long line widths reveal that the task of locating the beginning of successive lines of print is accomplished only with great difficulty. This is shown by the striking increase in regression frequency. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Article
With two groups of 20 college students as subjects, comparisons were made of the eye-movement pattern for reading 11 pt. vs. 8 pt. and 11 pt. vs. 6 pt. type, each presented with optimal length of line. Paragraphs from the Chapman-Cook Speed of Reading Test were used. Eye movements were photographed with the Minnesota eye-movement camera. Fixation frequency, words per fixation, pause duration, perception time, and regression frequency were recorded. The larger type showed some superiority in each instance, although the differences were significant beyond the 1% level only for pause duration and perception time. (PsycINFO Database Record (c) 2012 APA, all rights reserved)