The Galactic O-Star Spectroscopic Survey. I. Classification System and Bright Northern Stars in the Blue-Violet at R~2500

4.4 MB·English

by  A. Sota

The Galactic O-Star Spectroscopic Survey. I. Classification
System and Bright Northern Stars in the Blue-Violet at R∼25001
1 A. Sota2
1
0
2 J. Ma´ız Apella´niz2,3,4,5,6

n Instituto de Astrof´ısica de Andaluc´ıa-CSIC, Glorieta de la Astronom´ıa s/n, 18008 Granada, Spain
a
J

0 N. R. Walborn
2
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, USA

]
A
E. J. Alfaro
G
Instituto de Astrof´ısica de Andaluc´ıa-CSIC, Glorieta de la Astronom´ıa s/n, 18008 Granada, Spain
.
h
p
- R. H. Barb´a4
o
r Instituto de Ciencias Astron´omicas, de la Tierra y del Espacio, Casilla 467, 5400 San Juan, Argentina
t
s Departamento de F´ısica, Universidad de La Serena, Av. Cisternas 1200 Norte, La Serena, Chile
a
[

1 N. I. Morrell
v Las Campanas Observatory, Observatories of the Carnegie Institution of Washington, La Serena, Chile
2
0
0
R. C. Gamen
4
. Instituto de Astrof´ısica de La Plata (CCT La Plata-CONICET, Universidad Nacional de La Plata), Paseo
1
del Bosque s/n, 1900 La Plata, Argentina
0
1
1
J. I. Arias
:
v
Departamento de F´ısica, Universidad de La Serena, Av. Cisternas 1200 Norte, La Serena, Chile
i
X
r
a
ABSTRACT
We present the first installment of a massive spectroscopic survey of Galactic O stars, based
onnew, high signal-to-noiseratio, R∼2500digital observationsfrom both hemispheres selected
from the Galactic O-Star Catalog of Ma´ız Apell´aniz et al. (2004) and Sota et al. (2008). The
spectralclassificationsystemis rediscussedandanew atlasispresented,whichsupersedesprevi-
ousversions. Extensivesequencesofexceptionalobjectsaregiven,including types Ofc, ON/OC,
Onfp, Of?p, Oe, and double-lined spectroscopic binaries. The remaining normal spectra bring
◦
this first sample to 184 stars, which is close to complete to B = 8 and north of δ = −20 and
includes all of the northern objects in Ma´ız Apell´aniz et al. (2004) that are still classified as O
stars. The systematic and random accuracies of these classifications are substantially higher
than previously attainable, because of the quality, quantity, and homogeneity of the data and
analysis procedures. These results will enhance subsequent investigations in Galactic astronomy
and stellar astrophysics. In the future we will publish the rest of the survey, beginning with a
second paper that will include most of the south1ern stars in Ma´ız Apell´aniz et al. (2004).
Subject headings: binaries:general — stars:early type — stars:emission line,Be — stars:Wolf-Rayet —
surveys
1. Introduction than B = 13. Since then, we are deriving classifi-
cationsusingnew,uniformquality,highS/Nspec-
In Ma´ız Apell´aniz et al. (2004) we presented
trograms homogeneously processed and classified
the first version of the Galactic O-Star Catalog
according to well-defined standards. The survey
(GOSC), a collectionofspectralclassificationsfor
is described in Ma´ız Apell´aniz et al. (2010).
378GalacticO starsaccompaniedby astrometric,
How opportune and feasible is such a project?
photometric, group membership, and multiplicity
Ontheonehand,weareinabetterpositiontodo
information. Mostofthestarsinthatfirstversion
itthanwhensimilarsurveyswereattemptedinthe
had been classified by one of us (N.R.W.) two or
1960s and 1970s: there are more telescopes, bet-
three decades earlier using photographic spectro-
ter detectors, improved data reduction software,
grams. GOSC was subsequently expanded (ver-
and much larger reference databases. Further-
sion 2) by Sota et al. (2008), who added ∼ 1000
more, many of the targets are relatively bright,
stars that had at least one spectral classification
making the project accessible to 1-4 m class tele-
in the literature that identified them as O stars.
scopes. On the other hand, such a project still
Asaquicklookattheonlineversion1 ofGOSCv2
representsalargeandcomplicatedendeavor,with
reveals,thereisanunfortunatelylargedisparityin
the targets scattered along the Galactic Plane in
the literature spectral classifications for the stars
two hemispheres and requiring hundreds of obser-
there. Some of the discrepancies are due to dif-
vationnights. Also,since mostfields include none
ferentspectralresolutionsorsignal-to-noiseratios ′
or only a few additional O stars within ∼ 10 of
(S/N), others to variability in the stars (spectro-
theprimarytarget,theuseofafiberspectrograph
scopic binaries being the major culprit here), and
would be a waste of resources and a complication
still others to errors or different criteria among
forharmonizingthedatafromdifferentobservato-
classifiers. We believe it is important to correct
ries. Hence, the project is being conducted using
this situation, not only for the sake of the anal-
long-slit spectrographs.
ysis of individual stars but also because the use
TheearliestresultsfromGOSSSwerepresented
of inconsistent or incorrect spectral classifications
in a letter (Walborn et al. 2010b) that discussed
mayleadtoerrorsinthederivationofstatistically
thepresenceoftheCiiiλ4650blendinOfspectra.
based parameters such as the massive-star IMF
Inthisfirstpaperwepresent[a]anoverviewofthe
or the overall number of ionizing photons in the
project,[b]anatlasoftheblue-violetspectralclas-
Galaxy.
sification standards at R∼2500 from both hemi-
Thuswasbornin2007theideafortheGalactic
spheres that will be the basis of the rest of the
O-Star Spectroscopic Survey (GOSSS), a project
survey, and [c] a spectral library of 184 O stars
whoseprimarygoalis toobtainnewspectralclas-
without WR companions and with declinations
sifications of at least all Galactic O stars brighter ◦
larger than −20 . The majority of the stars in
this paper are from Ma´ız Apell´aniz et al. (2004);
1The spectroscopic data in this article were gathered
a few have been added to achieve completeness2
withthreefacilities: the1.5mtelescopeattheObservato-
rio de Sierra Nevada (OSN), the 3.5 m telescope at Calar to B =8.0, because of their presence in the same
Alto Observatory (CAHA), and the du Pont 2.5 m tele- slit as other O stars, or because of their inclusion
scope at Las Campanas Observatory (LCO). Some of the in Walborn et al. (2010b). The declination limit
supportingimagingdatawereobtainedwiththe2.2mtele-
is fixed by the accessibility from our northern ob-
scope at CAHA and the NASA/ESA Hubble Space Tele-
scope (HST). The rest were retrieved from the DSS2 and servatories but it turns out to be a useful value
2MASSsurveys. TheHSTdatawereobtainedattheSpace because it splits the numbers in the original cat-
Telescope Science Institute, which is operated by the As-
alog into two nearly equal parts. Paper II will be
sociation of Universities for Research in Astronomy, Inc.,
underNASAcontract NAS5-26555.
2VisitingAstronomer,CAHA,Spain. 2Asdescribedinthispaper,somestarspreviouslyclassified
3VisitingAstronomer,OSN,Spain. asB0VtoIIIhavebeenassignednewspectraltypesO9.7
V to III (previously, the O9.7 spectral type was defined
4VisitingAstronomer,LCO,Chile.
only for luminosity classes II to Ia). Since we have only
5e-mailcontact: [email protected]. observed a small fraction of the stars with B < 8.0 pre-
6Ramo´nyCajalfellow. viouslyclassified as B0, it is possiblethat we have missed
1http://gosc.iaa.es someO9.7starswithinthatmagnituderange.
2
the complement of part [c] of this one for declina- servatory (CAHA, Centro Astron´omico Hispano
◦
tions smaller than −20 . Future papers will ex- Alema´n). Also, since the image quality (see-
tend the O-starsample andthe wavelengthcover- ing+telescope+instrument) is usually better with
age; in both cases we already have abundant data TWINatCAHA thanwithAlbireoatOSN,some
taken3. We may also publish the spectrograms of of the bright northern stars with close compan-
the hundreds of non-O and low-mass stars (B4, ions were observedfrom CAHA in order to better
WR-includingWNhstars5 -,hotsubdwarfs)and spatially separate the two spectra.
the handful of O + WR systems that we are ob- The characteristics of the three setups are
taining as byproducts of our search. shown in Table 1. We used observations of the
same stars with two or three of the telescopes
2. Survey description to check the uniformity of the data9. The spec-
tral resolution of our OSN and LCO observa-
2.1. Blue-violetspectroscopywithR∼2500
tions as measured from the arc spectra turned
The primary goal of GOSSS is to obtain high out to be very similar and stable from night to
S/N (200-300) blue-violet spectrograms of all O night. R = 4500 ˚A/∆λ = 2500± 100 with
4500
stars with B < 13 at a high degree of uniformity ∆λ, the FWHM of the calibration lamp emission
and R ∼ 2500. Given those conditions, our first lines, being nearly constant over the full wave-
step was to select the telescopes and instruments lengthrangewithavalueof1.8˚A.ForourCAHA
with which to carry on the survey. For the north- data the spectral resolution was somewhat higher
ern part of the survey, we settled on the Albireo (R ∼ 3000, ∆λ ∼ 1.5 ˚A) and with a different
4500
spectrograph6atthe1.5mtelescopeoftheObser- dependence on wavelength. In order to provide
vatorio de Sierra Nevada (OSN), which can reach a uniform spectral library, a smoothing filter was
stars down to δ = −20◦. For the southern part applied to the CAHA data to achieve a constant
of the survey (δ < −20◦), we chose the Boller & ∆λ=1.8 ˚A for the full spectral range.
Chivens spectrograph7 at the 2.5 m du Pont tele- In this paper we present mostly OSN and
scope at Las Campanas Observatory (LCO). The CAHAdata,sincethemajorityofthe resultshere
du Pont telescope can reach the desired S/N val- correspond to the northern part of the survey.
ues for the dimmest stars in the sample within a Nevertheless, the atlas includes LCO data be-
reasonable total integration time (approximately cause for some spectral types southern standards
one hour), but in the north the 1.5 m at OSN re- are better than northern ones10. Our goal is to
quires significantly longer exposure times, which maintain our telescope triad for at least paper II
compromise the quality of the spectra due to the of the survey. If we eventually include new tele-
requiredinstrument stability. Therefore,the dim- scopes and/or instruments, we will first check for
mer stars (B > 11) in the northern part of the uniformity with the existing data.
survey were observed with the TWIN spectro-
The data in this paper were obtained between
graph8 at the 3.5 m telescope of Calar Alto Ob-
2007 and 2010. In some cases, observations were
repeateddue to focus andother instrument issues
3Wealsopointoutouttheexistenceoftworelatedprojects
detected after the fact. ForSB2 andSB3 spectro-
at higher spectral resolutions, OWN (southern hemi-
sphere, Barba´etal. 2010) and IACOB (northern hemi- scopic binaries, multiple epochs were obtained to
sphere, Sim´on-D´ıazetal. 2010), which are obtaining R ∼ observe the different orbit phases. In most cases
40000 optical spectrograms of hundreds of Galactic O,B, with known orbits, observations near quadrature
andWNstars.
were attempted.
4Among the stars we are classifying as early-B there are
some stars that had previously considered to be O stars, In order to reduce the large amount of data in
e.g. RY Sct and HD 194280. The latter, the propotype GOSSS, one of us (A.S.) wrote a pipeline in IDL.
late-OCsupergiant,hasbeenreclassifiedasBC0Iab.
5Hydrogen-richWNstarsappeartoberelativelyunevolved 9Note that from LCO it is possible to access declinations
verymassivestars(Crowtheretal.2010). muchfarther norththan δ=−20◦,thus providingalarge
6http://www.osn.iaa.es/Albireo/albireo.html .
overlapregionoftheskyforthethreeobservatories.
7http://www.lco.cl/telescopes-information/irenee-du-pont/ 10Forsomespectraltypestherearenonorthernstandardsin
telescopes-information/irenee-du-pont/instruments .
Ma´ızApella´nizetal.(2004).
8http://www.caha.es/pedraz/Twin/index.html .
3
2.00
1.75
HD 190 429 A+B
1.50
1.25
HD 190 429 A − O4 If
1.00
0.75
HD 190 429 B − O9.5 III
0.50
4600 4625 4650 4675 4700
Wavelength (Å)
Fig. 1.—[Topleft]False-colorrepresentationofaportionofaGOSSSlong-slitexposureofHD190429A+B.
The spectral direction is nearly parallel to the x axis. The bottom (brighter) component is A and the top
(weaker) component is B. [Top right] Spatial intensity cuts for two different wavelengths (one in red and
one in blue) for the data on the top left panel. The dotted lines show the two-component fit to the data.
[Bottom] Rectified extracted spectrum for each component and for the sum of the two. The continua are
all normalized to the value of the A component (∆B = 0.679 mag). Note the appearance of Ciii λ4647-
Ty
50-51 ˚A and Hei λ4713 absorptions and the change in the Heii λ4686 profile for the A+B spectrum when
compared to that of the A component.
4
Table 1: Telescopes, instruments, and settings used.
Telescope Spectrograph Grating Spectral scale Spatial scale Wav. range
(l/mm) (˚A/px) (′′/px) (˚A)
OSN 1.5 m Albireo 1800 0.66 0.85 3740−5090
LCO 2.5 m (du Pont) Boller & Chivens 1200 0.80 0.56 3900−5510
CAHA 3.5 m TWIN (blue arm) 1200 0.54 0.69 3930−5020
Thepipelinefirstappliesthebiasandflatandcal- 2.2. Complementary data, nomenclature,
culates a mask to eliminate cosmic rays and cos- and cataloguing
metic defects. Second, the data are calibrated in
Two problems that have complicated the spec-
wavelength and placed into the star rest frame.
tral classification of massive stars in the past are
Third,thestar(s)ineachlong-slitexposureis/are
[a] the presence of nearby resolved companions
identifed and extracted. Then, the spectra from
that may or may not contribute to the observed
different exposures (three or four per target) are
primary spectrum depending on the magnitude
combined and the final spectrogramis finally rec-
difference, separation,slit orientation, and seeing;
tified. The pipeline canbe runin either [a]a fully
and[b]themisidentificationofcomponentsinmul-
automated mode that is usually good enough for
tiple systems. Both issues are known to be the
a quick look at the telescope or [b] an interactive
sources of some discrepancies between literature
modethatallowsforthetweakingofsomeparam-
spectral classifications of the same target.
eterssuchasthemaskcalculationorthespectrum
In order to correct those two issues as much
rectification.
as possible, we used two strategies. On the
A special case is that of close pairs with small
one hand, we analyzed high-resolution imaging
magnitude differences (∆m). For those systems,
to identify and measure the magnitude differ-
we aligned the slit parallel to the line joining the
ences of nearby companions. For the northern
two stars to include both of them and we used a
part of the survey, this was done with Lucky-
custom-madeIDL fitting routinederivedfromthe
Imaging AstraLux observations at the 2.2 m tele-
MULTISPEC code (Ma´ız Apell´aniz 2005) to de-
scope of CAHA and HST imaging (GO pro-
convolve the two spatial profiles and extract the
grams 10602 and 11981, P.I. Ma´ız Apell´aniz,
spectra for the two stars. An example is shown
and archival data). The first results appeared
in Fig. 1. The procedure worksvery well for large
in Ma´ız Apell´aniz (2010) and will be used here.
separations but becomes increasingly harder for
For the southern part of the survey, we will
small values, especially if ∆m is large or the see-
use, among others, HST imaging from programs
ingisdegraded. Theclosestpairforwhichwewere
10205(P.I.Walborn),10602,and10898(P.I.Ma´ız
abletoextractseparatespectrathankstoexcellent
Apell´aniz). On the other hand, we searched the
seeing conditions was HD 17520 AB (∆m ≈ 0.7
′′ literature for results similar to those obtained
mag) with a separation of 0.316 (Ma´ız Apell´aniz
with AstraLux (e.g. McCaughrean & Stauffer
2010). Onthe other hand, we were unable to sep-
1994; Duchˆene et al. 2001; Mason et al. 1998;
arate σ Ori AB, which currently has a slightly
′′ Turner et al. 2008; Bouy et al. 2008; Mason et al.
lower separation(0.260)but a significantly larger
2009) and we plotted information from Simbad
∆m(≈1.6mag,Ma´ız Apell´aniz2010). Aswillbe
using Aladin images to ensure the correct identi-
shownlater, the use of such a deconvolutiontech-
fication of sources. In order to minimize possible
nique is the reason for the largest changes in the
future confusions,we providechartsfor somespe-
spectral classifications in this paper with respect
cific cases. We followed the component nomen-
to previous works.
clature of the Washington Double Star Catalog
The data from each observatory covers slightly
(Mason et al. 2001).
different wavelength ranges (Table 1). The spec-
Insomecases,theinformationderivedfromthe
trograms shown in this paper have been cut to
sources above allowed us to determine whether
show the same spectral range.
5
two or more visual components are spatially un- standardsweusedMGB11,anIDLcodedeveloped
resolved in our data. We considered that a sec- by one of us (J.M.A.) that overplots the two and
ondarycomponentiscapableofsignificantlymodi- allowstheusertoeasilychangefromonestandard
fyingthespectraltypeif|∆B|≤2.0. Insuchcases toanother. MGBalsoallowstheusertoartificially
weincludedinthenameofthestarthetwocompo- broaden the standard spectra to measure the line
nents (e.g. Pismis 24-1 AB or HD 93129 AaAb); broadening(see nextsection)andalsoto combine
forlargervaluesof|∆B|thesecondarycomponent twostandardspectraadjustingtheirvelocitiesand
was not included in the name. Note that when flux fraction in order to analyze spectroscopic bi-
we are able to spatially resolve a nearby compo- naries(seeFig.2for twoexamples). The software
nentandextractits spectrumindependently from wasindependentlyusedbytwoofthe authorsand
the primary, we do include the component name the results compared. In most cases there was an
in each case (e.g. HD 218195 A) even if |∆B| is excellent agreementin the classifications; discrep-
larger than 2.0. ancies were subsequently analyzed in more detail.
As previously mentioned, the GOSSS sample One important aspect is that spectral classifi-
was drawn from version 2.3 of GOSC (Sota et al. cation is subject to the effects of spectral, spatial,
2008). Our plans for the future include using the and temporal resolution as well as S/N. For ex-
new spectral classification to produce a new (3.0) ample, aSB2 may remainundetected without ad-
version of the catalog. That version will include equate resolution or temporal coverage, possibly
not only the spectral classifications but the spec- yielding anomalously wide lines due to blends; in
troscopic data themselves as well as the new dis- othercasessomeabsorptionlinesmaybetooweak
tances (Ma´ız Apell´aniz et al. 2008) derived from to be detected, e.g., Heii λ4542 at B0. In other
the new Hipparcos data reduction (van Leeuwen cases, a close visual binary may have historical
2007). composite spectra (hence, intermediate spectral
classificationsand/orpeculiarities)thatcannotbe
2.3. Spectral classification methodology separateduntilspatiallyresolvedspectroscopycan
be obtained. Such limitations are a major reason
Spectral classification according to the MK
for discrepantspectralclassificationsinthe litera-
process is carried out by [a] selecting a two-
ture. As previouslydescribed,we aretakingsteps
dimensional grid (in spectral type and luminos-
to minimize such effects (e.g., obtaining multiple-
ity class) of standard stars; [b] comparing the
epoch spectroscopy for known SB2s and to dis-
unknown spectrum with that grid, in terms of
cover new ones), but it is impossible to eliminate
the line ratios that define the different subtypes;
them completely. That is one of the reasons why
and[c]choosingthe standardspectrumthatmost
wepublishnotonlythespectraltypes,butalsothe
resembles the unknown spectrum, if appropriate
originalspectrograms,sincethatenablescompari-
noting any anomalies such as broad lines or dis-
sonwith pastorfuture results. Inthatregard,we
crepancies among different line ratios compared
havesearchedtheliteratureforspectrogramsthat
to the standards. The classificationcategoriesare
maybeinconflictwithourclassifications(because
discrete, whereas the phenomena are continuous,
of,e.g.,bettertemporalorspectralresolution)and
so interpolations or compromisesmay be required
analyzed those cases. We plan to continually up-
in some cases, which should be noted.
date the GOSC whenever new data justify it in
Manyofthe starsweselectedasstandardstars
the future.
for this paper have been previously used as such,
in some cases going back to the originaldefinition
3. Results
of the O subtypes. Nevertheless, in some cases
we noted inconsistencies that made us revise the This section constitutes the main body of this
spectralclassification,orwe found other,superior paper and is divided in three parts. First, we
definitions of the category among our expanded present the new atlas of standardO stars and the
sample, as detailed in the next section. For the associated spectral classification developments.
comparisonbetweentheunknownspectraandthe
11MarxistGhostBuster.
6
1.4
1.3
1.2
1.1
1.0
0.9
0.8
0.7
4000 4100 4200 4300 4400 4500 4600 4700 4800 4900
Wavelength (Å)
1.4
1.3
1.2
1.1
1.0
0.9
0.8
0.7
4000 4100 4200 4300 4400 4500 4600 4700 4800 4900
Wavelength (Å)
Fig. 2.— Two examples of spectral classifications of double-lined spectroscopic binaries using MGB. The
bottom(black)line showsthe spectrumtobe classifiedandthe top(redorgrey)linethe linearcombination
of the two standards. The flux fraction and velocity of each standard are indicated along with its spectral
type. [See the electronic version of the journal for a color version of this figure.]
7
Second, we briefly present the noteworthy char-
acteristics of some of the members of the pecu-
liar categories (Ofc, ON/OC, Onfp, Of?p, Oe,
SnsaBomn2-e+atwSlBaits3h)sttohaferstn)hoeirnmfutalhlleNOpoasrttpaherersr.ninFsitanhmaelplfylue,ll(waNetoldarstohatenhrdne 2.25 He I+II 3924 / 27Ca II is 3934 Ca II is 3968 He I 4009 He I+II 4026 C III 4068 / 69 / 70 N III 4097 He I 4121 He I 4144
sample.
2.00
HD 207 198 O9 II
3.1. Atlas and spectral classification sys-
1.75
tem developments
HD 16 429 A O9 II(n)
Ahistoricalandtechnicalreviewofthe current 1.50
spectralclassificationsystemforthe OBstarswas
HD 15 642 O9.5 II−IIInNwk
given by Walborn (2009a). Because of the un-
1.25
precedented quality and quantity of the present
dataset, several systemic developments and revi- HD 191 423 ON9 IInn
1.00
sions forthe O starsareintroducedin the present
work, which supersede previous procedures and
0.75
are described here. Classification standards are
launitsroltesaesistdpyisriconplvariTsedssaeeebsnsl,pteweed2ch,tiirnlaaenFlt-dthigyeuaplrneaetssetex4eqr–tue1fien2nvs;ceitvehaeseraenfitelrufiwsmxtesfiodpnueolrcustimfitrygai---l 00..2550 εH 3970 O II 4070 / 72 / 76 Si IV 4089 δH 4102 Si IV 4116
3950 4000 4050 4100 4150
classsequences atfixed spectraltypes (with a few Wavelength (Å)
exceptions because of positions unrepresented in
the current sample)12. This atlas replaces that of
Fig. 3.—Broadeningsequence[normal,(n),n,nn]
Walborn & Fitzpatrick (1990) for the O spectral
types. A list of qualifiers for O spectral types is for four O stars of similar spectral type. [See the
electronicversionofthejournalforacolorversion
provided in Table 3.
of this figure.]
Withregardtolinebroadening,wehaveconsis-
tently distinguished the three degrees (n), n, and
nn in this work. The ((n)) qualifier of Walborn
(1971) has not been applied, as it was judged too
marginal and close to the slight resolution differ-
ences among the different instruments involved.
Figure 3 shows the sequence from normal to nn
stars for stars around type O9 II. See Table 3 for Table 4: Spectral-type criteria at types O8.5–B0
the approximatevelocitiesthatcorrespondto(n), (comparisons between absorption-line pairs).
n, and nn, respectively13.
Spectral Heii λ4542/Hei λ4388 Siiiiλ4552/Heii λ4542
type and
Heii λ4200/Hei λ4144
O8 > N/A
O8.5 ≥ N/A
O9 = ≪
O9.5 ≤ <
O9.7a < ≤ to ≥
B0 ≪ ≫
a
Now used at all luminosity classes.
8
Table 2: Spectral classification standards.
V IV III II Ib Iab/I Ia
O2 HD93129AaAb
O3 HD 64568 ··· CygOB2-7
O3.5 HD 93128 Pismis 24-17 Pismis 24-1AB
O4 HD 46223 HD 168076AB HD15570
HD 96715 HD93250 HD16691
HD190429A
O4.5 HD15629 CygOB2-8C HD14947
HDE 303308 CygOB2-9
O5 HD 46150 HD 168112 CPD-472963
HDE 319699 HD93403
HD93843
O5.5 HD 93204 ··· CygOB2-11
O6 HD42088 HD101190 ··· HDE229196 ··· ··· HD 169582
HDE 303311
O6.5 HD 91572 HDE322417 HD190864 HD 157857 ··· ··· HD163758
HD12993 HD96946
HD152723
HD156738
O7 HD 93146 ··· CygOB2-4 HD94963 HD69464 ··· ···
HDE242926 HD151515 HD193514
HD 91824
HD 93222
15 MonAaAb
O7.5 HDE 319703A ··· HD163800 HD34656 HD17603 HD192639 ···
HD 152590 HD 171589 HD156154 9 Sge
O8 HD191978 HD97166 HDE319702 HD162978 BD-11 4586 HD225160 HD151804
HD 97848 λ OriA
O8.5 HD 46149 HD46966 HD114737 HD75211 HD125241 ··· HDE 303492
HD 57236 HD218195A
HD14633
O9 10Lac CPD-417733 HD24431 τ CMa 19Cep HD202124 αCam
HD216898 HD93028 HD93249 HD207198 HD148546
HD193443AB HD71304 HD152249
O9.5 AEAur HD192001 HD96264 δ OriAaAb HD76968 HD188209 ···
HD 46202 HD93027 HD154368
HD12323 HD155889 HD123008
HD96622
O9.7 υ Ori HD207538 HD189957 HD68450 V689Mon HD225146 HD195592
HD154643 HD152405 HD75222 HD 173010
HD10125 µNor HD105056
HD152424
B0 τ Sco ··· HD 48434 ··· ··· ··· ε Ori
HD122879
B0.2 HD2083 φ1 Ori HD6675 ··· ··· ··· ···
B0.5 HD 36960 ··· 1Cas ··· ··· ··· κ Ori
Notes Normal,italic,andboldtypefacesareusedforstarswithδ>+20◦,δ<−20◦,andtheequatorialin-betweenregion,respectively.
υ Ori,apreviousB0Vstandard,isnowanO9.7V.
τ Sco,apreviousB0.2Vstandard,isnowaB0V.
HD189957,apreviousO9.5IIIstandardisnowanO9.7III.
9
Table 3: Qualifiers used for spectral classification in this work and in others.
((f)) Weak Niii λ4634-40-42emission, strong Heii λ4686 absorption
(f) Medium Niii λ4634-40-42emission, neutral or weak Heii λ4686 absorption
f Strong Niii λ4634-40-42emission, Heii λ4686 emission above continuum
((f*)) Niv λ4058 emission ≥ Niii λ4640 emission, strong Heii λ4686 absorption (O2-3.5)
(f*) Niv λ4058 emission ≥ Niii λ4640 emission, weaker Heii λ4686 absorption (O2-3.5)
f* Niv λ4058 emission ≥ Niii λ4640 emission, Heii λ4686 emission (O2-3.5)
((fc)) As ((f)) plus Ciii λ4647-50-51emission equal to Niii λ4634
(fc) As (f) plus Ciii λ4647-50-51emission equal to Niii λ4634
fc As f plus Ciii λ4647-50-51emission equal to Niii λ4634
f?p Variable Ciii λ4647-50-51emission ≥ Niii λ4634-40-42at maximum; variable
sharp absorption, emission, and/or P Cygni features at H and Hei lines
((f+)) As ((f)) plus Siiv λ4089-4116emission (O4-8, obsolete, see subsection 3.1.3)
(f+) As (f) plus Siiv λ4089-4116emission (O4-8, obsolete, see subsection 3.1.3)
f+ As f plus Siiv λ4089-4116emission (O4-8, obsolete, see subsection 3.1.3)
(e) Probable Hα emission but no red spectrogramavailable
e Emission components in H lines
pe As e with emission components in Hei and/or continuum veiling
[e] Emission spectrum including Fe forbidden lines
e+ Feii and H emission lines (subcategories in Lesh 1968)
((n)) Broadened lines (not applied here, marginal)
(n) More broadened lines (vsini∼200 km/s)
n Even more broadened lines (vsini∼300 km/s)
nn Yet even more broadened lines (vsini∼400 km/s)
[n] H lines more broadened than He lines
nfp Heii centrally reversed emission, broadened absorption lines (Conti Oef)
N N absorption enhanced, C and O deficient
Nstr Moderate case of above (e.g. Niii λ4640 enhanced but not > Ciii λ4650)
C C absorption enhanced, N deficient
Nwk Moderate case of above
var Variation in line spectrum intensities or content
p Peculiar spectrum
z Heii λ4686 in absorption and > than both Hei λ4471 and Heii λ4542
10