Constraints on core-collapse supernova progenitors from correlations with H-alpha emission

Abstract

We present observational constraints on the nature of the different core-collapse supernova types through an investigation of the association of their explosion sites with recent star formation, as traced by H-alpha +[NII] line emission. We discuss results on the analysed data of the positions of 168 core-collapse supernovae with respect to the H-alpha emission within their host galaxies. From our analysis we find that overall the type II progenitor population does not trace the underlying star formation. Our results are consistent with a significant fraction of SNII arising from progenitor stars of less than 10 solar masses. We find that the supernovae of type Ib show a higher degree of association with HII regions than those of type II (without accurately tracing the emission), while the type Ic population accurately traces the H-alpha emission. This implies that the main core-collapse supernova types form a sequence of increasing progenitor mass, from the type II, to Ib and finally Ic. We find that the type IIn sub-class display a similar degree of association with the line emission to the overall SNII population, implying that at least the majority of these SNe do not arise from the most massive stars. We also find that the small number of SN `impostors' within our sample do not trace the star formation of their host galaxies, a result that would not be expected if these events arise from massive Luminous Blue Variable star progenitors. Comment: Accepted for publication in MNRAS

Full text

Page 1

arXiv:0809.0236v1 [astro-ph] 1 Sep 2008
Mon. Not. R. Astron. Soc. 000, 1–10 (2008) Printed 1 September 2008 (MN LATEX style file v2.2)
Constraints on core-collapse supernova progenitors from
correlations with Hα emission⋆
J.P. Anderson† and P.A. James
Astrophysics Research Institute, Liverpool John Moores University, Twelve Quays House, Egerton Wharf, Birkenhead CH41 1LD, UK
accepted 2008 August 14
ABSTRACT
We present observational constraints on the nature of the different core-collapse supernova
types through an investigation of the association of their explosion sites with recent star for-
mation, as traced by Hα+[NII] line emission. We discuss results on the analysed data of the
positions of 168 core-collapse supernovae with respect to the Hα emission within their host
galaxies.
From our analysis we find that overall the type II progenitor population does not trace the
underlying star formation. Our results are consistent with a significant fraction of SNII aris-
ing from progenitor stars of less than 10 M⊙. We find that the supernovae of type Ib show a
higher degree of association with HII regions than those of type II (without accurately tracing
the emission), while the type Ic population accurately traces the Hα emission. This implies
that the main core-collapse supernova types form a sequence of increasing progenitor mass,
from the type II, to Ib and finally Ic. We find that the type IIn sub-class display a similar
degree of association with the line emission to the overall SNII population, implying that at
least the majority of these SNe do not arise from the most massive stars. We also find that
the small number of SN ‘impostors’ within our sample do not trace the star formation of their
host galaxies, a result that would not be expectedif these events arise from massive Luminous
Blue Variable star progenitors.
Key words: stars: supernovae: general – galaxies: general – galaxies: statistics
1 INTRODUCTION
Despite years of observational and theoretical research on the na-
ture of supernova (SN) explosions and the properties of their pro-
genitors there remain substantial gaps in our knowledge of all SN
types. Although there are many different theoretical predictions
as to the nature of SN progenitors, the observational evidence to
discriminate between various progenitor scenarios remains sparse.
SNe can be split into two theoretical classes; SNIa which are
thought to arise from the thermonuclear explosion of an accret-
ing white dwarf, and core-collapse (CC) SNe which are believed
to signal the collapse of the cores of massive stars at the end points
in their stellar evolution.
Results from the first paper in this series (James & Anderson 2006,
JA06 henceforth) suggested that the SNIb/c arise from higher mass
⋆Based on observations made with the Isaac Newton Telescope operated
on the island of La Palma by the Isaac Newton Group in the Spanish Obser-
vatorio del Roque de los Muchachos of the Institute de Astrofisica de Ca-
narias, and on observations made with the Liverpool Telescope operated on
the island of La Palma by Liverpool John Moores University in the Spanish
Observatorio del Roque de los Muchachos of the Instituto de Astrofisica
de Canarias with financial support from the UK Science and Technology
Facilities Council.
† E-mail:jxa@astro.livjm.ac.uk
progenitors than SNII (albeit with small statistics; only 8 SNIb/c).
We test these initial results with an increased sample size enabling
us to distinguish between the various CC sub-types, and present
results from a combined sample of 100 SNII (that can be further
separated into 37 IIP, 8 IIL, 4 IIb, 12 IIn), 62 SNIb/c (22 Ib, 34
Ic and 6 that only have Ib/c classification), and 6 SN ‘impostors’.
We will present results and discussion of SNIa within the context
of the methods used in this paper elsewhere. We will also present
further research on the radial positions of SNe within galaxies, and
on correlations between CC SN type and local metallicity in fu-
ture publications. Here we concentrate on results on the progenitor
masses of the different CC SNe.
1.1Core-collapse supernovae
CC SNe are thought to be the final stage in the stellar evolu-
tion of stars with initial masses >8-10 M⊙, when fusion ceases
in the cores of their progenitors and they can no longer support
themselves against gravitational collapse. The different types of
CC SNe are classified according to the presence/absence of spec-
tral lines in their early time spectra, plus the shape of their light
curves. The first major classification comes from the presence of
strong hydrogen (H) emission in the SNII. SNIb and Ic lack any
detectable H emission, while the SNIc also lack the helium lines

Page 2

2
J.P. Anderson and P.A. James
seen in SNIb. SNII can also be separated in various sub-types.
SNIIP and IIL are classified in terms of the decline shape of their
light curves (Barbon et al. 1979; plateau in the former and lin-
ear in the latter), thought to indicate different masses of their H
envelopes prior to SN, while SNIIn show narrow emission lines
within their spectra (Schlegel 1990), thought to arise from inter-
action of the SN ejecta with a slow-moving circumstellar medium
(e.g. Chugai & Danziger 1994). SNIIb are thought to be intermedi-
ate objects between the SNII and Ib as at early times their spectra
are similar to SNII (prominent H lines), while at later times they
appear similar to SNIb (Filippenko et al. 1993).
Strong evidence has been presented to support the belief that SNII
and SNIb/c arise from massive progenitors, through their absence
in early type galaxies (van den Bergh et al. 2002), and the direct
detection of a small sample of progenitors on pre-explosion im-
ages (Smartt et al. 2004; Maund et al. 2005; Hendry et al. 2006;
Li et al. 2006; Gal-Yam et al. 2007; Li et al. 2007; Crockett et al.
2008). However, it is unclear how differences in the nature of
their progenitors produce the different SNe we see. It is clear that
there must be some process by which the progenitors of the dif-
ferent SNe lose part (or almost all in the case of SNIb and Ic) of
their envelopes prior to explosion. The differences in efficiency of
this mass loss process could be dependent primarily on progenitor
mass, with higher mass progenitors having higher mass loss rates
due to stronger stellar winds, and losing more of their envelopes.
In this picture a sequence of SNe types emerges from SNIIP and
IIL to SNIIb, SNIb and finally Ic having successively higher initial
masses. Thereare alsoother factors that probably play an important
role. Initialchemical abundance willalsoaffect theprogenitor mass
loss, with higher metallicity producing stronger radiatively driven
winds (e.g. Puls et al. 1996; Kudritzki & Puls 2000; Mokiem et al.
2007). It has also been proposed (Podsiadlowski et al. 1992) that
massive binaries could produce a significant fraction of CC SNe,
with mass transfer ejecting matter and leading to some of the vari-
ous CC sub-types.
Since the theoretical separation of SNe into two distinct explosion
classes by Hoyle & Fowler (1960), there have been many predic-
tions as to how the different CC SN types emerge from different
progenitors. There are two main theoretical routes to achieving the
observed different SN types. The first attempts to describe the full
range of SNe from a single star progenitor scenario. Heger et al.
(2003) and Eldridge & Tout (2004) produced SN progenitor maps
showing how variations in initial mass and metallicity produced
the different CC SN types. These models both predict that single
stars of up to ∼25-30 M⊙will produce SNIIP,with stars of slightly
higher mass producing SNIIL and IIb, and those of >30 M⊙ end-
ing their lives as SNIb/c (both authors also predict that these initial
mass ranges will shift to higher values with decreasing metallic-
ity). In both of these models no attempt was made to differenti-
ate between the SNIb and the SNIc, but one would presume that
within this single star scenario SNIc would arise from higher mass
progenitors than the SNIb as they have lost even more of their stel-
lar envelopes. Alternatively it could be that massive binaries pro-
duce the majority of CC SNe other than SNIIP (with these SNe
still arising from single star progenitors). The initial mass of the
stars producing SNIb/c, SNIIL and SNIIb would then be simi-
lar to those of SNIIP (12-20 M⊙, e.g. Shigeyama et al. 1990) but
would arise from binary evolutionary processes. There is also a
growing number of SNe that show evidence of binarity (e.g. SN
1987A; Podsiadlowski et al. 1990 and SN 1993J; Nomoto et al.
1993; Podsiadlowski et al. 1993; Maund et al. 2004). Recent com-
parisons of the observed ratio of SNIb/c rates to those of SNII
also argue that binaries are playing the dominant role in producing
SNIb/c (Kobulnicky & Fryer 2007), while Eldridge et al. (2008)
predict a SNIb/c rate produced by a combination of single and bi-
nary progenitors that best produces the observed SN rate. Again
one should note that these binary models group SNIb and Ic to-
gether and do not attempt to predict what differences in progenitor
produce these two types.
Given the different predictions for the origin of the CC SN types
described above, observations are needed to discriminate between
these models and thus firmly tie down the progenitors of the dif-
ferent SN types. However, apart from a small number of direct
detections of progenitors (Smartt et al. 2004; Maund et al. 2005;
Hendry et al. 2006; Li et al. 2006; Gal-Yam et al. 2007; Li et al.
2007; Crockett et al. 2008) this observational evidence remains
sparse. Therefore here we present results to test the above predic-
tions and constrain differences in progenitor mass of the different
CC SN sub-types by investigating the nature of their parent stellar
populations within host galaxies.
1.2Progenitor constraints from parent stellar populations
The most obvious way to determine the nature of SN progenitors is
to investigate the properties of their stars on pre-explosion images.
This has had some success although it is only possible for events
in very nearby galaxies and therefore the statistics remain low. An-
other way is to investigate how the rates of the various SN types
vary withdifferent parameters, such as redshift or host galaxy prop-
erties. Our approach is intermediate to these methods as we attempt
to constrain the nature of SN progenitors through investigating the
environments and stellar populations at the positions of historical
SNe. Here we concentrate on the association of the different CC
SNe types with recent star formation (SF) as traced by Hα emis-
sion.
Kennicutt (1998) states in a review paper on Hα imaging tech-
niques that: “only stars with masses >10 M⊙and lifetimes of <20
Myr contribute significantly to the ionising flux”. Thus, if our un-
derstanding of this lineemission iscorrect, we can use thisassump-
tion as a starting point to constrain the relative stellar lifetimes and
therefore the relativemasses of thevarious SN progenitors, through
investigating how accurately the different SN types trace the emis-
sion. In JA06 we presented a statistic to quantitatively measure
the association of individual SNe with the Hα emission of their
host galaxies, and presented results from an initial galaxy sample
(HαGS,discussed in §2). It was found that overall theSNII progen-
itor population did not trace the underlying SF of their host galax-
ies, with a significant fraction lying on regions of low or zero emis-
sion line flux which were ascribed to a ‘Runaway’ fraction of pro-
genitor stars (however, this assumed that SNII arise from progeni-
torsof >10 M⊙).TheSNIb/cdidappear tofollow theemissionim-
plying that these progenitors come from higher mass stars than the
SNII, although the statistics on this class were small (only 8 SNe
for SNIb and Ic combined). This SN/galaxy sample has now been
significantly increased, enabling the full parameter space of CC SN
progenitors to be investigated and results from this increased sam-
ple are presented here.
The paper is arranged as follows: in the next section we present
the data and discuss the reduction techniques employed, in §3 we
summarise thestatisticintroduced inJA06 and used throughout this
paper, in §4 we present the results for the different CC SN types, in
§5 we discuss possible explanations for these results and their im-
plications for the relative masses of the SN progenitors, and finally
in §6 we draw our conclusions.

Page 3

Constraints on CC SN progenitors
3
2 DATA
The initial galaxy sample that formed the data set for JA06 was
the Hα Galaxy Survey (HαGS). This survey was a study of the
SF properties of the local Universe using Hα imaging of a repre-
sentative sample of nearby galaxies, details of which can be found
in James et al. (2004). 63 SNe (of all types, including SNIa) were
found to have occurred in the 327 HαGS galaxies through search-
ing the International Astronomical Union (IAU) database1.
Through three observing runs onthe Isaac NewtonTelescope (INT)
and an on-going time allocation with the Liverpool Telescope (LT)
we have now obtained Hα imaging for the host galaxies of 133
additional CC SNe, the analysis of which is presented here. The
LT is a fully robotic 2m telescope operated remotely by Liverpool
John Moores University. To obtain our imaging we used RATcam
together with the narrow Hα and the broad-band Sloan r’ filters.
Images were binned 2×2 to give us 0.278′′size pixels, and the
width of the Hα filter enabled us to image target galaxies out to
∼2400 kms−1. The INT observations used the Wide Field Camera
(WFC) together with the Harris R-band filter, plus the rest frame
narrow Hα (filter 197) and the redshifted Hα (227) filters enabling
us to image host galaxies out to ∼6000 kms−1. During our 2005
INT observing run we also used the SII filter (212) as a redshifted
Hα filter and imaged 12 SN hosting galaxies at distances of ∼7500
kms−1. The pixel scale on all INT images is 0.333′′per pixel and
with both the LT and INT our exposure times were ∼800 sec in Hα
and ∼300 sec in R.
These additional SNe/galaxies were chosen from the Padova-
Asiago SN catalogue2, as specific CC SN types were more com-
plete for the listed SNe. At a later date all SN type classifications
taken from the Padova-Asiago catalogue were checked through a
thorough search of the literature and IAU circulars, as classifica-
tionscan often change after the initialdiscovery and therefore those
in the catalogue may not be completely accurate. The full list of
SN types is given in Appendix B, where references are given if
classifications were changed from those in the above catalogue.
The main discrepancies were the classification of the so called SN
‘impostors’ as SNIIn in the Padova-Asiago catalogue. These are
transient objects that are believed to be the outbursts from very
massive Luminous Blue Variable stars (LBVs), which do not fully
destroy the progenitor star and are therefore not classed as true
SNe(e.g. van Dyk et al. 2000; Maund et al. 2006). Six such objects
were found in our sample, and the results on these ‘impostors’ are
presented and discussed separately in the following sections.
The distance limit for our sample (mainly set from the available
Hα filters during observing runs) enables us to resolve the stellar
population close to the SN position, and we also exclude edge on
galaxies because of extinction effects and increased projection un-
certainties. We do not include results on SNe where images were
obtained within 18 months for SNII and a year for SNIb/c after the
catalogued explosion epoch. This is to ensure that our images are
not contaminated with residual SN light and that the Hα emission
that we detect is due to the underlying HII regions and not asso-
ciated with the SNe themselves. Through the above telescope time
allocations we have therefore obtained data on host galaxies of al-
most all discovered CC SNe (that have been classified IIP, IIL, IIb,
IIn, Ib, and Ic) that meet our selection criteria and were observable
within the Hα filters of the two telescopes.
There are obvious biases within a set of data chosen in the above
1http://cfa-www.harvard.edu/iau/lists/Supernovae.html
2http://web.pd.astro.it/supern/
way. As we use any discovered SNe for our sample, the various
different biases in the different SN surveys that discovered them
mean that the galaxy/SN sample is by no means representative of
the overall SN populations. Bright, well studied galaxies will be
over represented, as will brighter SNe events that are more easily
detectable. However, firstly we are not analysing the overall host
galaxy properties (as we will show when discussing the statistics
we use in §3), but are analysing where within the distribution of
stellar populations of the host galaxy the SNe are occurring. Sec-
ondly, the small number of CC sub-types that are discovered means
that no individual survey can currently manage to analyse the prop-
erties of their host galaxies or parent stellar populations in any sta-
tistically significant way (most statistical observational studies do
not even attempt to separate the Ib and Ic SN types). Taking our
approach enables us to make statistical constraints on all the major
CC SN sub-types. The results that are presented in this paper are on
the analysis of the parent stellar populations of 100 SNII, of which
37 are IIP, 8 IIL, 4 IIb and 12 IIn, 6 SN ‘impostors’, plus 22 Ib,
34 Ic and 6 that only have Ib/c as their classification, from both the
initial HαGS sample and our additional data described above.
2.1Data reduction and astrometric methods
For each SN host galaxy we obtained Hα+[NII] narrow band
imaging, plus R- or r’-band imaging used for continuum sub-
traction. Standard data reduction (flat-fields, bias subtraction
etc) were achieved through the automated pipeline of the LT
(Steele et al. 2004), and the INT data were processed through the
INT Wide Field Camera (WFC), Cambridge Astronomical Survey
Unit (CASU) reduction pipeline. Continuum subtraction was then
achieved by scaling the broad-band image fluxes to those of the
Hα images using stars matched on each image, and subtracting the
broad-band image from the Hα images. Our reduction made use of
various Starlink packages.
Thenext process wastoobtainaccuratepositions forthesitesof our
SNe on their host galaxy images. This astrometric calibration was
achieved by transferring the accurate astrometry of XDSS second
generation Palomar SkySurvey images3, onto matching galaxy im-
ages in our sample (the full process is described in JA06). In nearly
all cases astrometric calibration was achieved with fit residuals of
<0.2′′. Withaccurate positions obtained for the SNe sites we could
now analyse to what degree the different SNe were associated with
the distribution of Hα emission within their host galaxies.
In figures 1 and 2 we show two examples of Hα images of the host
galaxies of SNe from our sample, with SN positions derived from
the above astrometric calibration. We intend to present all our Hα
and R-band imaging of SN host galaxies in a future publication
(Ivory et al. 2008, in prep), where we will release all of our data
for public use along with host galaxy derived characteristics such
as SF rates and Hα equivalent widths.
3PIXEL STATISTICS
Previous works investigating the association of SNe with HII re-
gions within host galaxies (e.g. Bartunov et al. 1994; van Dyk et al.
1996)havegenerally usedsomemeasureof thedistancetothenear-
est bright HII region to quantify the association of each individual
SN to the high mass SF within their host galaxies. However, this
3downloaded from http://cadcwww.dao.nrc.ca/cadcbin/getdss

Page 4

4
J.P. Anderson and P.A. James
Figure 1. An example negative continuum subtracted Hα image from our
data (image taken with the WFC on the INT); SN 2001ac (SN ‘impostor’)
(position indicated by circle/arrow), within the host galaxy NGC 3504. The
scale bar is 20′′. The NCR value for this SN is 0.000.
Figure 2. Another example negative continuum subtracted Hα image
(WFC, INT); SN 2004bm (Ic) (position indicated by circle/arrow), within
the host galaxy NGC 3437. The scale bar is 20′′. The NCR value for this
SN is 0.704.
brings various problems when defining the nearest bright HII re-
gion and therefore the distance to measure. In JA06 we presented
a quantitative statistic that reduced any ambiguity in the measure-
ments of each SN, by analysing where the count of the SN hosting
pixel falls within the overall distribution of Hα pixel values of the
galaxy. Theexactdetailsof how thisstatisticisformedcanbefound
in JA06, and here we summarise this process and the main points
on how this can be used in analysing the associations of the differ-
ent SN types to the emission.
Thepixelsinthecontinuum subtracted Hαimageswerefirstbinned
3×3 to reduce the pixel-to-pixel noise level and enable us to deter-
mine the SN-containing pixel with a degree of certainty. The pixels
were then sorted into increasing pixel count. The cumulative distri-
bution of these values was then formed and normalised, with neg-
ative values set to zero, giving a normalised cumulative rank pixel
value function (NCR henceforth) running from 0 to 1, with one en-
try for each pixel on the host galaxy image. Within this distribution
therefore, values of 0 correspond to zero emission line flux or sky
values, whereas a value of 1 corresponds to the centre of the bright-
est HII region on the image. Figures 1 and 2 illustrate the use of
this statistic, with the SN ‘impostor’ 2001ac falling away from any
detected Hα emission in Fig. 1 and therefore having an NCR value
of 0.000, whereas in Fig. 2 the SNIc 2004bm falls on a bright HII
region and therefore has a high NCR value of 0.704.
When we form the NCR it is found that the majority of values ly-
ing above the sky level within this distribution are small and indi-
vidually contribute little to the overall flux, but by force of num-
bers they do contribute a significant amount to the underlying SF.
Alongside this, there will be relatively few NCR values close to 1,
but those that are will individually make a significant contribution
to the overall flux. Thus the distribution is formed so that if a SN
progenitor population is drawn from the same stellar population
that produces the Hα flux, one would expect a mean NCR value
for that SN type of 0.5 and a flat distribution. This is therefore the
initial hypothesis that we work from, that if the progenitors of CC
SNe trace the same high mass SF as does Hα emission, we expect
their NCR values to form a flat distribution. We can then investi-
gate whether there are any differences in the mean NCR values and
distributions of the different CC SN sub-types and what this may
imply for differences in the relative lifetimes and masses of their
progenitors.
A full discussion of the errors associated with this statistic was pre-
sented in JA06, therefore here we will summarise the main errors;
those presented with the results are the statistical errors found on
the various distributions. The most obvious error is that associated
with the determination of the SN containing pixel. This was inves-
tigated by determining the NCR value of each SN for a 3×3 pixel
box centred on the SN pixel (meaning that after already binning
3×3 we sample regions ∼2.5′′and 3′′on the LT and INT images
respectively). A comparison was then made of the median NCR
value of the box with the SN pixel. This was repeated for the new
sample where we find the size of the errors to be consistent with
those from JA06, and there are in general no significant differences
between the SN pixel NCR values and those of the median value
of the surrounding pixels. For the overall SNII NCR distribution
we find a mean difference of 0.027 between the NCR value of the
SN pixel and the median pixel. The rms difference in NCR value
is 0.163 where, as in JA06 this is dominated by around five cases
wherethereisasignificant differencebetween thevalues. However,
overall the NCR analysis seems to give results which are robust to
positional errors of 1-2′′. In JA06 possible errors due to the adopted
sky level were investigated but these were found to be insignificant.
Finally a Monte Carlo analysis was performed on the effects of
pixel-to-pixel noise on the NCR value. Again this effect was found
to be small, with errors appearing to be random and producing no
tendency to bias the results in any particular direction. We will now
present the results formed from using the above described statistic
on the various CC SN types.

Page 5

Constraints on CC SN progenitors
5
Figure 3. Histogram of the NCR values of the SN-containing pixels within
the cumulative Hα flux distribution for the 100 SNII
4 RESULTS
4.1 SNII
Figure 3 shows the overall distribution of NCR values for the 100
SNII our sample. It is immediately clear that the positions of SNII
do not follow the overall distribution of SF as traced by the Hα line
emission, confirming the result of JA06. In fact there isan excess of
∼35% of SNII that fall on sites of little or zero Hα flux compared
to what would be expected if these SNe followed the distribution
of Hα emission. The probability of the SNII progenitor population
being drawn from a flat distribution (i.e. following the line emis-
sion), calculated using a Kolmogorov-Smirnov (KS) test is <1%.
Overall the mean NCR value for SNII is 0.252 with a standard er-
ror on the mean of 0.027. We will now present the results obtained
when separating the SNII into their various sub-types. It should be
noted here that ∼40% of our type II SNe do not have designated
sub-types and are only classified as SNII.
4.1.1 SNIIP
SNIIP are the most abundant SNII sub-type observed and therefore
it is not surprising that these are the most abundant of those with
sub-type classification in our sample. It is also to be expected that
their distribution of NCR values follows that of the overall SNII
population as can be seen when comparing Figs. 3 and 4, with a
KS test showing that the two distributions (SNe classified as IIP
removed from the overall II distribution) are formally consistent
with each other. Again, if one assumes that the majority of those
unclassified SNII will be of type IIP (i.e. if sufficient data were
available on their light curves etc), this is to be expected. The mean
NCR value for the SNIIP population is 0.263 (0.048).
4.1.2SNIIL
The 8 SNIIL population have a mean NCR value of 0.255 (0.112)
and seem to follow the same distribution as the overall SNII popu-
lation.
Figure 4. Histogram of the NCR values of the SN-containing pixels within
the cumulative Hα flux distribution for the 37 SNIIP
Figure 5. Histogram of the NCR values of the SN-containing pixels within
the cumulative Hα flux distribution of the 12 SNIIn
4.1.3 SNIIb
The 4 SNIIb have a mean NCR value of 0.460 (0.162), higher than
that of the overall SNII population. To measure the significance
of this difference we used a Monte Carlo analysis. Removing the
SNIIb from the distribution of SNII NCR values we calculated the
fraction of times that a mean NCR value of ?0.460 (SNIIb mean
value) was produced when four values were drawn at random from
the overall SNII distribution. We found that there is only a ∼6%
chance that the SNIIb parent population is drawn from the same
distribution as that of the rest of the SNII.
4.1.4 SNIIn
Figure 5 shows the distribution of the NCR values for the 12 SNIIn.
The mean NCR value for this SN type is 0.256 (0.088), and these
SNe seem to follow the same stellar population as that of the over-
all SNII population. Using a KS test we find that there is only ∼1%
chance that these SNe are drawn from a flat distribution (i.e. fol-
lowing the distribution of Hα emission).