astro-ph/9509060 11 Sep 1995 STATISTICS Send o print requests to The distribution of H Introduction are still not frequentlyhigh found in the literature star formation parameters acrossused galactic in discs programmes However aimed atregions provide determining an global observational massive base whichreview is by increasingly Kennicutt cent Complete massive catalogues star of formation in these spiral galaxies see e g the Abstract Received accepted July th A the ble morphological type For threewith values galaxies published we for determined other spiral galaxies of compara subtracted M Rozas NGC and NGCand interarm zones these separately We nd that for NGC I of the H individual H determined the p ositions angular sizes NGC and uxes NGC of their NGC and NGC We have tions for all the H luminosity functions arm and interarm zones for each galaxy The slop es of the the spiral NGC NGC NGC NGC galaxies those Key ing pro cesses NGC and discuss implications for massive star form NGC Your NGC thesaurus co des NGC are will NGC b e inserted byA A hand manuscript later no preetdeD epartement Physique Universit ede Montr eal C P SuccursaleInstituto Centre de Ville H C Astrof sicade Montr eal Qu ebec Canarias J Canada E La Laguna Tenerife Canarias Spain LUMINOSITY slop es errors SAMPLE quality words found i i 1 region samples We construct luminosity func of the of We J E Beckman H i i from catalogues determination regions and describ e statistical prop erties ISM present images luminosity functions for i i earlier regions in the disc and separately for i i for H J E Beckman regions is an excellent tracer of re and i i the new OF work of regions AND detailed 1 complete We the FUNCTIONS and J H Knap en high sp eci cally GRAND compare slop es grand design quality galaxies statistical sample PROPERTIES are the our for equal continuum spiral arm agree individual results M analyses galaxies within 2 DESIGN well and to were analysis of the cal NGC grand design spiral galaxies in the H that the H ranging in morphological type from Sb to Irr They found more H have studied the H cutt Edgar Ho dge for hereafter Sa referred galaxies to we asphological KEH refer types to have b een Caldwelltype studied et spiral For al galaxies example although for other Kenni LFs galaxies of other mor function with of H arm for the interarm is signi cantly ve di erent galaxies from in that their foris sample the shallower for KEH which constructed the H slop es nd any signi cant di erence in the arm and interarm LF lished a catalogue of the H of tribution of H M by Rand hereafter referredto to that as obtained R The from dis di erence an analysis b etween of arm the and interarm H LFs a result similar uncertainties of the tsslop es to of the the slop es armet for and al that interarm galaxy hereafter LFs referred are to as equal KACB within shows the that the NGC p er same program as the galaxies analysed in the present pa NGC In This is the rst of two pap ers in which we describ e an In the present Cepa Beckman have studied the distribution The distribution i i are observed regions in the barred spiral NGC and did not The i i H similar in regions OF i i as GALAXIES same authors NGC region LF can b e describ ed image i i used N as luminosity functions LFs regions in the disc of NGC Knap en of L that part than early type pap er we will discuss i i H H quality to by region LFs in a sample of galaxies / i i used galaxy they of our i i L KACB regions and 2 Cepa REGIONS by i i 0 NGC regions in the inner 5 ongoing program to the Late type spirals p ossess arm and interarm LFs for which R in images presented the galaxies and did Beckman and ASTROPHYSICS discs emission line forms ASTRONOMY nd exclusively late by a p ower law These i i the and geometri of AND region LF in a i i part pub signi cant NGC region LF image their galaxies IN of image 0 now the LF of 0
M Rozas et al Statistics and Prop erties of H i i regions I
Table Basic prop erties of galaxies
Galaxy Arm Class Type R D R A Dec
0
RC Mp c
h m s 0 00
NGC SABb c rs
h m s 0 00
NGC Sc s
h m s 0 00
NGC SBb c s
h m s 0 00
NGC SABb c rs
We can therefore compare our results directly with those is describ ed in Sec The results are presented in terms
found in R KACB and more qualitatively with KEH of LFs in Sec and brie y discussed in terms of star
In Table we give basic prop erties of the four galaxies formation pro cesses in Sec The main conclusions are
presented in this pap er NGC NGC NGC summarized in Sec
and NGC Column two in the Table is the arm class
as de ned by Elmegreen Elmegreen where class
Observations and data reduction
galaxies have the b est de ned spiral arms Column
three is the morphological type de Vaucouleurs et al
The observations of NGC NGC and NGC
hereafter RC Column four is the radius at mag
th th
were made during the nights of September and
2
arcsec R in arcmin from the RC Column ve is
25
with the m William Herschel Telescope WHT on
the distance in Mp c as determined from the redshift as
La Palma We used the TAURUS instrument in imaging
1 1
suming H km s Mp c Finally columns six and
0
mo de as a re imaging camera mounted at the Cassegrain
seven give the p ositions of the centres of the galaxies for
fo cus of the telescop e The detector used was an EEV
ep o ch from RC
00 00
CCD with pro jected pixel size Observ
The galaxies were selected to give a balance b etween
00
ing conditions were very go o d with seeing FWHM
barred and non barred ob jects with the aim of comparing
as measured in the nal images and photometric sky We
and contrasting their prop erties
obtained two exp osures of seconds each p er galaxy
NGC is an angularly small weakly barred galaxy
one through a A wide lter whose central wavelength
with well de ned spiral arms The nuclear region is rel
coincided with that of the redshifted H emission from the
atively small and it is noted for several dusty regions
galaxy A for NGC A for NGC and
which can b e clearly recognized on a blue image not
A for NGC and another through a nonred
repro duced here
shifted H lter A with A width for continuum
NGC is a late type spiral galaxy of type Sc It
subtraction NGC was observed during the night of
th
shows no sign of a bar and has no obvious companion
May using the same telescop e and instrumen
nearby It has two main spiral arms and some smaller
tation Given the angular size of this galaxy we decided to
arm sections as describ ed by Bo eshaar and Ho dge
take three sets of line and continuum images of s each
who studied the distribution of H i i regions in
comp osed of separate s images at three di erent
this galaxy
p ositions of the galaxy disc We used the A wide A
NGC is an angularly small barred galaxy of arm
lter for the line images and the equally wide A l
class It has two spiral arms and very little emission
ter for the continuum imaging H continuum subtracted
b etween them The brightest star forming regions are
images were pro duced for these three p ositions separately
at the ends of the bar The nucleus of NGC is a
see b elow which were then pasted together into one nal
LINER
H image of the galaxy The resolution of the nal image
00
NGC is the second mixed type weakly barred
was limited to the resolution of the sub image with
spiral in our sample A bright and well de ned ring
the p o orest resolution which also led us to re bin the nal
00 00
structure surrounds the very bright nuclear region
image to a pixelsize Photometric calibration
which is elongated along an axis at nearly right angles
was achieved by comparison of the nal image with an
to the bar axis We will not discuss this inner region in
image of the inner part of the disc of NGC obtained
the present pap er NGC hosts a Seyfert nucleus
earlier with the WHT
Bo er Schulz
Standard reduction routines were used following
In Fig we show grey scale representations of the closely the pro cedure used in KACB for NGC A
continuum subtracted H images of NGC NGC bias level was rst subtracted and the images were then
NGC and NGC The observations and data re corrected using appropriate dawn sky at elds Next a
duction pro cedures are summarized in Section and the value for the sky background was obtained by measuring
H i i region selection and the construction of the catalogues the background and asso ciated noise levels in p ortions of
M Rozas et al Statistics and Prop erties of H i i regions I
Fig Grey scale representations of the H continuum subtracted images of NGC a NGC b NGC c and
NGC d
the images where there was no emission from the galaxies Pro duction of the H i i region catalogues
In order to align the two line and continuum images of
each galaxy p ositions of several foreground stars were de
termined making gaussian ts and the o band image was
displaced to coincide with the H image using the stellar
Table The r m s noise in instrumental counts column
p ositions The accuracy of the sup erp ositions was in all of the background subtracted H images the lower limit to
the luminosity of the detected H i i regions in each galaxy col
cases b etter than pixels In the case of NGC we
and the approximate size of the smallest catalogued regions
also used foreground star p ositions to align the three sub
col
images for subsequent combination into the nal galaxy
image
Galaxy r m s log L erg s D p c
H
NGC
Once the aligned images were cleaned of cosmic ray
NGC
e ects see KACB for a discussion and an o ccasional bad
NGC
pixel the continuum image was subtracted from the line
NGC
H continuum image giving nett H ux As we ex
p ected see KACB the b est scaling factor was was close
to unity for the subtraction of the continuum image from
the line image given that b oth images were taken through
Before pro ducing the H i i region catalogues we agged
lters with very similar bandwidths and transmissions and
the foreground stars in the images These are distinguish
in practically the same observing conditions We made
able from H i i regions by their regular circular shap es in
sure the scaling factor used cannot b e in error by
the original unsubtracted images and b ecause they show
more than by exp erimentally scaling the continuum
more intensity in the continuum than in the corresp ond
by factors less than unity In that case residual
ing H continuum subtracted image Ideally foreground
continuum was left in the nuclear and arm regions after
stars should not show up at all in the H continuum
subtraction If factors greater than unity were employed
subtracted images but in most cases some residual is seen
negative intensities app eared in these same re
due to e g di erences in p oint spread function alignment
gions As describ ed in KACB we prefer to use the present
or stellar emission b etween the line and continuum im
metho d over one where foreground stars which may not
ages or b ecause the star is saturated on one or b oth im
have a at sp ectrum in the range considered are used to
ages Emission in the H image coincident with a fore
scale the images
ground star on the continuum image was considered resid
ual starlight and not entered in the catalogue as an H i i
region As a selection criterion for H i i regions we imp osed
that a feature must contain at least nine contiguous pix
Table Calibration constants H luminosity which corre
els each with an intensity of at least three times the r m s
sp onds to a single instrumental count
noise level of the lo cal background Any ob ject not meet
ing this criterion was treated as indistinguishable from
Galaxy L erg s count
H
noise In Table we list the r m s noise of the background
NGC
NGC
subtracted H images the lower limit to the luminosities
NGC
of the detected H i i regions in each galaxy as well as the
NGC
sizes of the smallest catalogued regions the last two quan
tities are derived directly from the adopted selection cri
terion
In identifying and quantifying the parameters of the
Absolute ux calibration was carried out using obser H i i regions we had to overcome three complicating ef
vations of standard stars from the lists of Oke fects Firstly many H i i regions app ear to overlap on the
Stone and Filipp enko Greenstein The image Without attempting to analyze what fraction of
luminosity in H which corresp onds to a single instru these overlaps implies real contact and what fraction is
mental count in each of our images is given in Table for merely a pro jection e ect we adopted the solution pro
each galaxy p osed in R and followed in KACB of counting each p eak
M Rozas et al Statistics and Prop erties of H i i regions I
NGC i PA NGC in H as representing a single H i i region The ux of each
NGC NGC We also de H i i region was then estimated by integrating over the pix
termined the diameter and the H luminosity in erg s els which could b e reasonably attributed to a given region
of each H i i region and assigned an index letter to sp ec One will undoubtedly miss a number of H i i regions that
ify whether the given region is in an arm or an interarm are to o weak to b e detected in the vicinity of stronger
zone The catalogues are available via the CDS or from emitters close by This will in uence the lower end of the
the authors In Fig we show schematically the p ositions LF but is not a signi cant factor in the determination of
of the H i i regions in the disc of NGC NGC the shap e of the true LF at the higher luminosity end
NGC and NGC on a depro jected RA dec grid R Secondly an H i i region is not necessarily circular
centred on the nucleus of each galaxy Arm and interarm We adopted as an e ective radius the mean of the maxi
regions are distinguished by using di erent symbol styles mum and minimum radii as measured for each H i i region
Thirdly the presence of di use H may lead to ill de ned
edges of H i i regions introducing some systematic errors
Luminosity functions
ab ove all for the weakest regions We have not tried to
Total LFs
quantify this e ect but note its p ossible in uence
After identifying each H i i region we measured the p o
The luminosities in H of the regions catalogued range
36 40 1
sition of its centre with resp ect to the centre of the galaxy
from some to erg s The regions at the lower
and derived a mean radius We determined the ux of each
end of the luminosity range are in all probability ionized
region using a programme which was previously used by
by a single star so that the LF at these levels essentially
KEH R and KACB that integrates counts within a cir
traces the stellar LF although in this range the sample
cular ap erture A constant sky value was subtracted from
loses its completeness Most of the regions detected have
37 39 1
the images b efore determining the integrated uxes see
luminosities b etween and erg s which cor
49
ab ove but we made additional corrections for a small
resp onds to Lyman continuum luminosities b etween
51
number of H i i regions due to lo cally varying background
and ionizing photons p er second The upp er limit to
levels We estimate that sky subtraction errors cannot in
the number of ionizing photons p er second from a single
49
uence the luminosity determinations by more than ab out
star is of order thus the ma jority of the regions de
in even the weakest regions
tected must b e ionized by the radiation from groups of
stars The brightest regions may b e physically di erent
from the others e g formed under di erent physical con
Table Total number of catalogued H i i regions for each
ditions or p ossibly the result of mergers
galaxy for arm and interarm separately and for the whole
galaxy
Table LF p eaks column and slop es col and
Galaxy Arm Interarm Total
lower luminosity limits for completeness col L in erg s
NGC
NGC
Galaxy LF p eak LF slop e completeness limit
NGC
NGC log L
NGC
NGC log L
NGC log L
NGC log L
NGC log L
Most previous studies of H i i regions in spiral galaxies
M l og L
were made using relatively low resolution imaging For ex
ample in an early study of H i i regions in NGC Bo e
shaar Ho dge identi ed only regions Ho dge
Kennicutt identi ed only regions in NGC The LFs are presented in Fig where we have plot
We identify regions in NGC and in ted the decimal logarithm of the number of H i i regions
NGC which re ects the improvement in resolution
in bins of width dex along the ordinate and the log
1
and sensitivity in the present data There are no previ
arithm of the luminosity in erg s along the abscissa
ously published H i i region catalogues for NGC or
The cuto levels for each galaxy are a result of our crite
NGC Table lists the number of catalogued H i i re rion for discriminating H i i regions from the background
gions in each galaxy for the disc as a whole and for arm noise they are visible as a drop in the observed LFs to
and interarm zones separately
the low luminosity side although the LFs in this region
For all the H i i regions catalogued for each galaxy we are also in uenced by incompleteness The p ositions of
determined equatorial co ordinate o sets from the nucleus the p eaks of the LFs are in go o d agreement with those
and depro jected distances to the centre in arcsec using determined by R for M and by KACB for NGC
the inclination angles and p osition angles given by Grosb l indicating that the data sets used for these studies and the
M Rozas et al Statistics and Prop erties of H i i regions I
Fig Depro jected maps of the program galaxies comp osed of schematic representations of the p ositions of the measured HI I
regions Op en symbols represent HI I regions in the spiral arms lled symbols HI I regions in the interarm disc R A and Dec
o sets are in arcsec relative to the centre of the galaxy a NGC upp er left b NGC upp er right c NGC
lower left and d NGC lower right
present one are comparable in quality In Table we show present pap er In Table we give the two sets of slop es
the p eaks of the LFs of the galaxies as well as the com and the limits of their resp ective luminosity ranges Again
pleteness limits b elow which faint H i i regions esp ecially we note that the lower range slop es are very close to each
in the spiral arm zones may have b een missed These were other in value as are also the upp er range slop es imply
de ned by departures from uniformity in slop e of the ob ing common underlying physical b ehaviour from galaxy to
served functions galaxy
Omitting the centres of the galaxies and those H i i re
gions b elow the completeness limits in luminosity we t
Arm and Interarm LFs
ted functions of the form N L AL dL to the data In
In Fig we show the LFs for the H i i regions of the arm
all cases the slop es lie within the range
and the interarm regions separately for all the galaxies Of
in agreement with KEH Table lists the slop es of the
the galaxies analyzed NGC do es not show particu
LFs and the lower luminosity limits of the ts for the four
larly well de ned arms in H which made a classi cation
galaxies in the present pap er For comparison we also list
as interarm or arm H i i regions di cult In all galaxies
NGC from KACB and M from R Note that
we classi ed as interarm regions those not obviously con
to b e consistent with the literature we add to the
nected to the nearest spiral arm We checked the results
slop es of the plotted LFs see KACB
we obtained this way by comparing the p ositions of the
An interesting characteristic of all the LFs is a blip
spiral arms as seen in the continuum image where the
in the otherwise monotonically falling linear function close
38 6 1
edges of the arms are b etter de ned and reclassi ed a
to L erg s We rep orted this phenomenon in
few H i i regions In general the H i i regions lying in the
NGC KACB and it is also seen in M R
interarm zones are far fewer in number than those in the
The blip is accompanied by a change in slop e with a
arms as is evident from Table
steep er gradient at higher luminosities rep orted previ
The slop es of the LFs of the interarm regions of
ously by KEH They discuss this change in terms of a
NGC NGC and NGC are equal to those
transition from normal to sup ergiant H i i regions but this
of the arms within the range of uncertainties inherent in
is a phenomenological rather than a physical description
the measurements The interarm LF for NGC has
The discrepancy in the LFs would then indicate that some
no statistical signi cance since the number of interarm
well de ned physical mechanism causes the formation of
regions is very small
such sup ergiant H i i regions One p ossibility is that the
The slop es of the LFs for the arms and the interarm
blip and the change of slop e are due to the change
zones for each galaxy and the limits to the luminosities for
from ionization b ounded to density bounded regions and
which they were determined app ear in Table The ta
marks the limitation on the size of the neutral gas clouds
ble also incorp orates NGC KACB and M R
which give birth to the stellar asso ciations The change
1
of slop e shows up very clearly at log L erg s in Due to the smaller numbers of H i i regions the LFs of the
interarm regions are displaced vertically with resp ect to
NGC and NGC in NGC and NGC it
1
the arm LFs and the p eaks are displaced horizontally to
is just a shade lower log L erg s and log L
1
lower luminosities
erg s resp ectively although in NGC the break is
less sharp than in the others It will b e of great interest to
see whether on theoretical grounds we should exp ect the
Implications for star formation
break to o ccur at constant luminosity in which case these
di erences of order dex in the luminosity of the break
The most signi cant result of this part of the study of
p oint would corresp ond to di erences in the assumed dis the distribution of H i i regions in the discs of NGC
tances of order NGC NGC and NGC is the determina
In Fig we show the LFs of the galaxies but now with tion of slop es of the LFs not only in the disc as a whole
two ts in two luminosity ranges ab ove and b elow the but also in the arms and the interarms zones separately
break p oint in the LF noted ab ove This change in slop e Within the errors of determination we nd that the arm
referred to previously by KEH for some of the galaxies and interarm LFs have the same slop es for NGC
they analyzed is clearly present in all the galaxies of the NGC and NGC For NGC there is little
M Rozas et al Statistics and Prop erties of H i i regions I
Fig HI I region LFs of NGC a upp er left NGC b upp er right NGC c middle left NGC d middle right NGC e lower left from KACB and M f lower right from R Lines indicate b est ts see text
M Rozas et al Statistics and Prop erties of H i i regions I
Table LF slop es and luminosity limits for the two luminosity ranges in erg s
NGC NGC NGC NGC
range slop e range slop e range slop e range slop e
l og L l og L l og L l og L
l og L l og L l og L l og L
Table Slop es of arm and interarm LFs and corresp onding lower luminosity limits in erg s
Galaxy Arm Interarm
slop e of LF Luminosity limit slop e of LF Luminosity limit
NGC log L log L
NGC log L log L
NGC log L log L
NGC log L log L
NGC l og L l og L
M log L log L
statistical signi cance to the slop e of the interarm func the arm and interarm LFs in NGC NGC and
tion NGC are the same
Comparing the results of this pap er with those pub
The spiral arms of NGC NGC and
lished in the literature we see that in several grand design
NGC are all well de ned The ma jority of the H i i
spirals the slop e of the arm LFs are shallower than the
regions are lo cated in the arms as are al l the most lu
inter arm LFs but that this is certainly not a uniform
minous regions KEH measured the H i i regions of a set
picture It is interesting to note that for the three galax
of galaxies with well de ned spiral arms separating the
ies in this pap er for which we were able to determine
arm from the interarm zones and found as a general re
arm and interarm LF slop es we cannot distinguish these
sult that the slop es are di erent due to the presence of a
slop es within the errors of determination Comparing only
larger number of high luminosity H i i regions in the arms
the more recently published data of go o d and comparable
than in the interarm zones This result is however not
quality we nd that only M R has signi cantly dif
necessarily reliable since the e ect of H i i region crowd
fering arm and interarm LF slop es whereas in NGC
ing in the arms may have a ected the luminosity limit in
KACB NGC NGC and NGC this pa
the photographic data of KEH which have relatively p o or
p er one cannot distinguish arm from interarm LF slop es
resolution and S N ratio more so than in the more recent
M is a galaxy with an exceptionally well developed and
studies including the present one and esp ecially in the
strong spiral system where the high gas surface densities
arms
combined with strong density wave compression in the
Cepa Beckman found that the slop es of the
arms may lead to a shallower molecular cloud mass sp ec
arm and interarm LFs are the same for NGC but did
trum and a shallower arm H i i region LF as suggested by
nd a signi cant di erence in NGC Cepa Beck
R In the other galaxies the arms may not b e strong
man Rand attributed the di erence he ob
enough to make the LF there di er signi cantly from the
served in M b etween arm and interarm LF to a less
LF in the interarm discs
steep mass sp ectrum of molecular clouds in the arms i e
to a higher prop ortion of massive clouds giving rise to
relatively more H i i regions of higher luminosity It is rea
Conclusions
sonable to assume that the coalescence of clouds in the
The main results of the present pap er where we ana
arms is due to the enhanced global gas density and to the
lyze the LFs of the H i i regions in the discs of NGC
greater frequency of cloud cloud interactions there
NGC NGC and NGC are summarized
Finally KACB found that in NGC the slop es of
b elow
the arm and interarm LFs are the same within the uncer
tainties The authors stated that in NGC the spiral From high quality continuum subtracted H images
arms might not b e strong enough to build up the large of the grand design spirals NGC NGC
cloud masses needed to pro duce the number of giant HI I NGC and NGC we have catalogued a to
regions that make the arm LF shallower than the inter tal of and H i i regions resp ectively
arm LF This could well b e the reason that the slop es of The catalogues include p ositions radii and uxes of all
M Rozas et al Statistics and Prop erties of H i i regions I
Fig HI I region LFs of NGC a upp er left NGC b upp er right NGC c lower left and NGC d
lower right Lines indicate b est ts for two luminosity ranges The slop es are clearly di erent for the lower and upp er luminosity
ranges
H i i regions Tables containing all these data are avail of these galaxies but equal in others This di erence
able through CDS or directly from the authors may b e due to the presence of su ciently strong arms
The LFs for the whole discs total LFs have slop es to change the arm LF in those grand design galaxies
that agree well with slop es for other galaxies of com such as M with strong arms where coherent density
parable morphological types wave e ects are of imp ortance
The LFs for the di erent subsamples of H i i regions
Acknowledgements We acknowledge Mr L Gentet for his ex
have the same slop es in the arms and in the interarm
tensive help in the pro duction of the H i i region catalogue for
regions within the limits of the measurement uncer
NGC We thank Dr A F J Mo at and Dr R F Peletier
tainties
for helpful comments on the manuscript The William Her
Comparing the LF slop es for arm and interarm
schel Telescope is op erated on the island of La Palma by the
regions as determined for NGC NGC
Royal Greenwich Observatory in the Spanish Observatorio del
NGC and NGC with results for other
Ro que de los Muchachos of the Instituto de Astrof sicade Ca
grand design galaxies in the literature esp ecially M
narias This work was partially supp orted by the Spanish DG
and NGC we nd that they are di erent in some
ICYT Direcci on General de Investigaci on Cient ca y T ec
M Rozas et al Statistics and Prop erties of H i i regions I
Fig Arm and interarm H i i region LFs of NGC a NGC b NGC c NGC d NGC e from KACB and M f from R Drawn lines indicate b est ts
M Rozas et al Statistics and Prop erties of H i i regions I
nica Grant No PB This research has made use of
the NASA IPAC Extragalactic Database NED which is op
erated by the Jet Propulsion Lab oratory California Institute
of Technology under contract with the National Aeronautics
and Space Administration
References
Bo er B Schulz H A A
Bo eshaar G O Ho dge P W ApJ
Caldwell N Kennicutt R C Phillip s A C Schommer
R A ApJ
Cepa J Beckman J E A AS
Cepa J Beckman J E A AS
de Vaucouleurs G de Vaucouleurs A Corwin H G Buta
R J Paturel G Fouqu e P Third Reference
Catalogue of Bright Galaxies RC Springer New
York
Elmegreen D M Elmegreen B G ApJ
Filipp enko A V Greenstein J L PASP
Grosb l P J A AS
Kennicutt R C in Star formation in stel lar systems
Eds G Tenorio Tagle et al Cambridge Univ Press
Cambridge
Kennicutt R C Edgar B K Ho dge P W ApJ
KEH
Knap en J H Arnth Jensen N Cepa J Beckman J E
AJ KACB
Oke J B ApJS
Rand R J AJ R
Stone R P S ApJ
This article was pro cessed by the author using Springer Verlag
a
L T X A A style le L AA version E