1: \documentclass[preprint2]{aastex}

2: %\documentclass[12pt]{emulateapj}

3: 

4: %\usepackage{latexsym,amssymb}	% Additional symbols

5: %\usepackage[dvips]{epsfig}		% Postscript figures

6: %\usepackage{amsmath, amssymb}              % for mathematical symbols etc

7: 

8: \newcommand {\nh}{N$_{2}$H$^{+}$\ }

9: \newcommand {\nhns}{N$_{2}$H$^{+}$}	%nh symbol w no space afterwards.

10: \newcommand {\co}{C$^{18}$O\ }

11: \newcommand {\cons}{C$^{18}$O}		%co symbol w no space afterwards.

12: \defcitealias{Kirk06}{Paper~I}

13: %\newcommand {\hk}{\bf}

14: \newcommand{\hk}{}

15: 

16: \begin{document}

17: \title{Dynamics of Dense Cores in the Perseus Molecular Cloud}

18: 

19: %\slugcomment{Version 4.0, \today}

20: 

21: \author{Helen Kirk\altaffilmark{1,2}, Doug Johnstone\altaffilmark{1,2}, and Mario 

22: 	Tafalla\altaffilmark{3}}

23: \altaffiltext{1}{Department of Physics \& Astronomy, University of Victoria, 

24: 	Victoria, BC, V8P 1A1, Canada; hkirk@uvastro.phys.uvic.ca}

25: \altaffiltext{2}{National Research Council of Canada, Herzberg Institute of 

26: 	Astrophysics, 5071 West Saanich Road, Victoria, BC, V9E 2E7, 

27: 	Canada}

28: \altaffiltext{3}{Observatorio Astron\'omico Nacional (IGN), Alfonso XII 3,

29: 	E-28014 Madrid, Spain}

30: 

31: %%%%%%%%%%%%%%%%%%%%%%%%%%%

32: \begin{abstract}

33: We survey the kinematics of over one hundred and fifty candidate (and 

34: potentially star-forming) dense cores in the Perseus molecular cloud 

35: with pointed \nhns(1-0)

36: and simultaneous \cons(2-1) observations.  Our detection rate of

37: \nh is 62\%, rising to 84\% for JCMT SCUBA-selected targets. 

38: In agreement with previous observations, we find that

39: the dense \nh targets tend to display nearly thermal linewidths, particularly

40: those which appear to be starless (using Spitzer data), indicating

41: turbulent support on the small scales of molecular clouds is minimal.

42: For those \nh targets which have an associated SCUBA dense core,

43: we find their internal motions are more than sufficient to provide support 

44: against the gravitational force on the cores.  

45: Comparison of the \nh integrated intensity

46: and SCUBA flux reveals fractional \nh abundances between $10^{-10}$ and

47: $10^{-9}$.

48: We demonstrate

49: that the relative motion of the dense \nh gas and the surrounding

50: \co gas is less than the sound speed 

51: in the vast majority of cases ($\sim$90\%).  The point-to-point

52: motions we observe within larger extinction regions appear to be

53: insufficient to provide support against gravity, although we sparsely

54: sample these regions.  

55: \end{abstract}

56: %%%%%%%%%%%%%%%%%%%%%%%

57: \keywords{infrared: ISM: continuum -- ISM: individual (Perseus) -- 

58: 	ISM: structure -- stars: formation -- submillimetre}

59: 

60: 

61: %%%%%%%%%%%%%%%%%%%%%%

62: \section{INTRODUCTION}

63: Stars form in the densest regions of a hierarchy of structures that exist 

64: within a molecular cloud.  Supersonic motions dominate on the largest 

65: scales \citep[e.g.,][]{Larson81} but appear to be much reduced on the 

66: smallest scale, that of a preprotostellar core \citep[e.g.,][]{Benson89}.  

67: Understanding what physical processes are at play at each scale ranging

68: from the largest to the smallest is a challenge that both observers

69: and theorists face.  The challenge is complicated

70: by the fact that each molecular cloud possesses different properties --

71: local environment has a strong effect on the star formation process.

72: Some molecular clouds, such as Taurus, display isolated star-forming 

73: cores which are quiescent and have a low star formation efficiency,

74: while others such as Orion display clustered

75: star-forming cores and have more turbulent motions and a higher

76: star formation efficiency \citep[e.g.,][]{Cohen79}.  While clustering increases

77: the complexity of a region and hence the difficulty in understanding

78: what processes are at play, the majority of stars appear to form

79: in environments which are clustered to some degree, and hence it

80: is important to study such regions to understand the impact of clustering.

81: In this paper, we examine the Perseus molecular cloud -- a cloud which

82: is less clustered and confused than structure in the Orion molecular cloud, 

83: but is not as isolated and quiescent as the Taurus 

84: molecular cloud.  The Perseus molecular cloud consists of a chain of 

85: distinct, well-known small clustered environments 

86: in which stars are forming -- e.g., NGC1333 and IC348 \citep{Bally_handbook}.  

87: There is a wealth of data uniformly spanning this cloud, courtesy of

88: the Spitzer `c2d' Survey \citep{Evans03} and the COMPLETE (Co-Ordinated

89: Molecular Probe Line, Extinction, and Thermal Emission) Survey

90: \citep[see][for a summary of the publicly available data]{Ridge06}.

91: The continuum data from the latter survey (SCUBA thermal dust emission

92: and 2MASS near-IR extinction) allow for the determination of

93: the (column) density structure of the cloud.  In particular, 

94: we \citep[][hereafter Paper I]{Kirk06} characterized the environments 

95: in which dense cores (which could eventually evolve to form a star) 

96: were themselves able to form.  In doing so, we determined a set of 

97: constraints on

98: the column density structure of the cloud which simulations should

99: match.  The goal of the present work is to extend that set of density

100: constraints to a set of dynamical constraints.

101: 

102: To sample the behaviour of the dense gas in cores across the molecular

103: cloud, we performed pointed \nh observations on a set of locations

104: including candidate dense cores from submillimetre data, points of 

105: high visual extinction from the Palomar plates, and peaks of large-scale

106: extinction from 2MASS data.  With the addition of the Spitzer 

107: data \citep{Jorgensen07},

108: we are able to differentiate between unevolved dense starless cores

109: and their more evolved protostellar bretheren.  We couple {\hk these} data

110: with the existing continuum data discussed above in order to determine

111: the dynamical behaviour of the dense gas within the cloud on a variety

112: of scales.  Our data {\hk do} not provide the resolution or full sampling

113: of some other recent surveys \citep[e.g.,][]{Walsh06}, but {\hk have}

114: the advantage of providing (sparse) sampling across a much larger

115: area, and {\hk are} thus complimentary to these other works.

116: 

117: %%%%%%%%%%%%%%%%%%%%%%%%%%%

118: \section{SOURCE CATALOG}

119: 

120: \subsection{SCUBA submillimetre}

121: The Perseus molecular cloud has been mapped over roughly 

122: $\sim 3.5$ square degrees in the submillimetre (sensitive to thermal 

123: radiation by dust grains)

124: at 850~$\mu$m with the Submillimetre Common User Bolometer Array (SCUBA) at

125: the James Clerk Maxwell Telescope in Hawaii. 

126: In \citetalias{Kirk06}, we identified approximately fifty cores

127: in the SCUBA data using a 6\arcsec\

128: sampled map of all existing data in the cloud.  The majority of {\hk these} 

129: data {\hk were} archival \citep[see][]{Hatchell05}, but also included some `fast scan'

130: maps (with insufficient integration time to allow for the typical 3\arcsec\ 

131: sampling) taken as part of the COMPLETE project \citep{Ridge06}. 

132: \citetalias{Kirk06} compared the properties of the SCUBA cores 

133: to the environment

134: they inhabit through comparison to near IR extinction data with 2.5\arcmin\ 

135: resolution (see discussion of {\hk these} data below), hence the consistent

136: areal coverage was of greater importance than resolution.  All of

137: the submillimetre cores were

138: found in regions of previously identified star-formation (e.g., B1, NGC1333), 

139: and none were found in the fast-scan mapped regions.  For our current 

140: project, we utilize maps created with a finer 3\arcsec\ sampling which

141: allows for better separation of close cores and determination of core

142: properties such as radius.  Appendix A discusses the re-reduction of

143: the data we performed, and presents a full source catalog.  We identify

144: seventy two submillimetre cores above our usual S/N level 

145: \citep[these cores are also used in][]{Jorgensen07}, as well as 15

146: additional potential submillimetre cores which do not satisfy our

147: S/N criteria.

148: 

149: The original 6\arcsec\ sampled map formed the basis of our target list of

150: dense cores to observe in \nhns(1-0).  We supplemented this with a

151: list of potential SCUBA cores which fell below the detection

152: threshold of our core-identification procedure in order to have as 

153: complete a list as possible.  This resulted in 89 dense core

154: targets.  These target positions are listed in Table~\ref{targets}

155: along with the associated better-defined SCUBA core from this paper.

156: 

157: \subsubsection{Spitzer \& young protostars}

158: The Perseus molecular cloud was also one of the clouds surveyed by the

159: Spitzer `c2d' project \citep{Evans03}.  While the submillimetre

160: wavelengths covered by SCUBA are sensitive to the dusty envelopes

161: of both starless cores and enshrouded protostars, the IR wavelengths 

162: probed by Spitzer, particularly the shorter wavelengths, are sensitive

163: to the central accreting object and hence young protostars.  Thus Spitzer

164: observations are ideal in distinguishing between starless cores and

165: their more evolved bretheren. 

166: The full Spitzer dataset is described in \citet{Jorgensen06} and

167: \citet{Rebull07}, while a 

168: catalogue identifying young protostellar candidates using a combination of

169: Spitzer and SCUBA data is presented in \citet{Jorgensen07}.  Note that the

170: work of \citet{Jorgensen07} utilizes the 3\arcsec\ sampled SCUBA core list

171: included in Appendix~A.

172: Table~\ref{targets} also shows whether the targets are associated with

173: a young protostar as detected with Spitzer.

174: 

175: A second comprehensive catalog listing the protostellar and 

176: starless cores in Perseus has also recently been published by 

177: \citet{Hatchell07} where the sources were classified using SEDs fit to

178: IRAS, Spitzer, SCUBA, and Bolocam observations.  A significant

179: difference between the two catalogs is that the Hatchell catalog

180: used only the shorter wavelength Spitzer IRAC data, while the 

181: J\o rgensen catalog relied primarily on the longer wavelength

182: Spitzer MIPS data.

183: The bulk of the two 

184: protostellar catalogs agree, but the J\o rgensen catalog 

185: contains five protostars not in the Hatchell catalog

186: and the Hatchell

187: catalog contains seventeen protostars not in the J\o rgensen

188: catalog.  

189: 

190: Of the five sources only in the J\o rgensen catalog,

191: one was outside the region included in the Hatchell catalog,

192: two were classified as starless cores in the Hatchell catalog, and 

193: the remaining two had submillimetre emission below the

194: threshold for identification.  The J\o rgensen catalog had

195: three criteria for classification as a protostar -- a set of

196: colour criteria for a source across the IRAC (3.6, 4.5, 5.8~$\mu$m) 

197: and MIPS (24~$\mu$m) bands, spatial

198: coincidence of a MIPS 24~$\mu$m source with a SCUBA core, or detection

199: of a SCUBA core with a high central concentration (see Appendix~A

200: for the definition of concentration).  None of the

201: seventeen protostars only identified

202: in the Hatchell catalog are associated with a high submillimetre

203: central concentration, therefore inclusion in the J\o rgensen catalog

204: would require a detection in the MIPS 24~$\mu$m band plus fulfilment

205: of either

206: the colour criteria or SCUBA core association.  We searched the Spitzer data 

207: \citep{Rebull07,Jorgensen06,Evans03} around each of the

208: seventeen protostars only in the Hatchell catalog.  Twelve of the 

209: seventeen did not have MIPS detections with a signal to noise above

210: five (seven of these had close SCUBA cores while five did not).

211: Of the remaining five, none had close SCUBA cores or colours satisfying

212: the colour criteria of the J\o rgensen catalog.

213: Of the sources only in the J\o rgensen catalog, we observed four of 

214: the five and all had

215: detections (our sources \#21, 22, 74, and 95).  Of the sources only 

216: in the Hatchell catalog, we observed ten of the seventeen and had detections

217: for seven (our sources \#92, 94, 101, 109, 110, 121, and 150). 

218: 

219: A range of linewidths, centroid velocities, and peak

220: intensities were found for these sources, therefore any potential

221: mis-classification is unlikely to bias our results. 

222: 

223: \subsection{Extinction - 2MASS}

224: From the COMPLETE Survey, we also have a map of the total column density 

225: created from extinction

226: measures derived using the NICER technique \citep{Lombardi01}

227: on the 2MASS dataset \citep{Ridge06,Alves}.  

228: This technique utilizes near infrared reddening of background stars in 

229: three wavelength bands (two colour indices) in order to determine 

230: the total column density of dust.  The

231: resolution of this map is 2.5\arcmin\ and spans the range of A$_{V}= $3 - 11 

232: within the region mapped by SCUBA (Figures 2 and 3 of \citetalias{Kirk06}).

233: \citetalias{Kirk06} analyzed {\hk these} extinction data and identified structures 

234: in the column  density on two scales.  We 

235: utilized the maxima in the smaller-scale structures (the `extinction cores' 

236: of \citetalias{Kirk06}) to

237: identify a further 24 targets for our \nh survey.  In later sections

238: of this paper, we utilize the larger-scale structures (`extinction supercores'

239: of \citetalias{Kirk06}) to

240: define the environments in which the dense cores inhabit.  The extinction 

241: target positions,

242: along with the associated information from the extinction and submillimetre

243: maps are given in Table~\ref{targets}.

244: 

245: \subsection{Palomar Plates}

246: The final extension to the dense core candidate target list was taken from

247: visually-selected targets in the red POSS-II Palomar plates.  We focussed

248: on plates in regions devoid of SCUBA cores in order to maximize the

249: extent of our coverage of environments within the cloud.  This provided

250: an additional 44 targets.  The information on these targets is given

251: in Table~\ref{targets}.

252: 

253: Figure \ref{fig_targets} shows all of the targets we selected overlaid on the

254: extinction map of the Perseus molecular cloud.

255: 

256: \section{OBSERVATIONS AND DATA REDUCTION}

257: We made pointed observations of the 157 targets

258: using the 30~m IRAM telescope in Pico Veleta, Spain.  We observed

259: \nhns(1-0) in both polarizations using the (AB) 100~GHz receivers and

260: simultaneously observed \cons(2-1) in both polarizations using the (CD) 230~GHz

261: receivers, in both cases utilizing frequency switching.  We used the 

262: VESPA correlator and smoothed to 0.05~km/s channels.  The beamsize is

263: $\sim$25\arcsec\ for \nh and $\sim$11\arcsec\ for \cons.

264: Each observation was made for 2 minutes and had an rms of T$_{A}^{*}\sim$0.1~K.

265: We made multiple pointings on some of the cores to better resolve spectral

266: features or search for a signal to a higher sensitivity.

267: We made four-point maps (offsets of 25\arcsec\ in RA and dec around the 

268: central pointing)

269: around some of the targets in order to distinguish extended structures 

270: and search

271: for cores offset from the assumed position.  In the interests of minimizing

272: observing time, this was done for only a subsample (62) of the cores.

273: 

274: We detected signal in \nh in  62\%  of our targets at the central

275: position, with 

276: the rate rising to 84\%  for the SCUBA-selected targets. 

277: Since \nh is a `late time' {\hk molecular ion}, not attaining significant

278: abundance until after $\sim10^{5}$~yrs \citep{Aikawa03}, this implies 

279: most of the SCUBA sources are at least this old.

280: 

281: Virtually every target had a \co detection (96\%).  All but one of the

282: spectra with signal in \nh had a corresponding detection in \cons.

283: Examination of the one rogue case revealed that higher than average 

284: noise in the \co spectrum was likely responsible for the lack of a detection.  

285: 

286: Table~\ref{table_detect}

287: shows the full break-down of the detection rates and numbers for the various 

288: target selection methods.  There are several factors which likely go

289: into the vast difference in success rates for the different selection

290: criteria.  Both the 2MASS and Palomar plate-selected targets have greater

291: uncertainties attached to a potential core's position than the SCUBA-selected

292: cores.  The 2MASS map has a resolution of 2.5\arcmin\ and thus is 

293: insensitive to the small density peak of an individual core, rather, it 

294: represents a smoothed average of any large- plus small-scale dense 

295: structures along the line of sight.  The Palomar plates

296: have better resolution (1\arcsec) but are only sensitive to a few 

297: magnitudes of visual extinction.  Our detection efficiency for this

298: portion of the survey is similar

299: to the NH$_{3}$ survey of \citet{Benson89} which used the original Palomar

300: plates to identify candidate cores.

301: 

302: \nh has a critical density of $\sim 10^{5}$~cm$^{-3}$ \citep{Tafalla02} and thus

303: is sensitive to only the densest gas within molecular clouds.

304: The SCUBA-selected and Palomar-selected targets were all chosen on the

305: basis of apparent (column) density enhancements which would indicate

306: the site of a dense core.  While this was not the case for the

307: 2MASS-selected targets (the resolution allowed only for the identification

308: of peaks in the large-scale structure), those 2MASS-selected targets

309: which had detections all lie on or close to structure visible in the

310: SCUBA observations.  Hence all of the targets where we detected \nh

311: are likely to be dense cores rather than less dense gas unassociated with

312: any small-scale structure.

313: 

314: \subsection{Fitting the Spectra}

315: We reduced the \nh and \co data from IRAM using CLASS \footnote{See 

316: {\tt http://www.iram.fr/IRAMFR/GILDAS}.}.

317: First, we fit a baseline to each spectrum individually using a 4th

318: order polynomial.  The resultant spectra were then folded and summed

319: where multiple pointings existed.

320: 

321: To fit the seven components of the \nhns(1-0) spectra, we used CLASS's 

322: HyperFine Split fitting routine, with the relative frequencies and

323: optical depths for \nh taken from \citet{Caselli95} and 

324: a frequency of 93176.258~MHz for the \nhns(JF$_{1}$F=101-012) `isolated'

325: component \citep{Lee01}. 

326: 

327: In some cases (20 \nh spectra), two separate components were clearly required

328: for a good fit.  We interpret these spectra as belonging to two separate

329: entities, rather than a central dip caused by self-absorption.  

330: Appendix~B discusses the evidence for this interpretation {\hk of} these cores.

331: 

332: Table~\ref{table_fits_nh} shows the best-fit line parameters

333: found using CLASS for the \nh spectra -- the centroid velocity, 

334: velocity dispersion (measured in terms of FWHM), 

335: the total optical depth, the baseline level (alternatively, the standard 

336: deviation in regions with no line emission), and the line rms 

337: (alternatively the standard

338: deviation in the fitting residuals where there is line emission).

339: All fits were visually inspected;

340: those of poor reliability due to low S/N are noted in the final

341: column -- these have reasonable centroid velocities

342: but poor dispersions since noise limits the determination of the

343: extent of the line.  We include the

344: less secure fits only in the analysis of centroid velocities.

345: The integrated intensities were also measured for each spectrum

346: using the tdv function{\footnote {\hk The tdv function,

347: a part of the spectral cube package, calculates the integrated intensity 

348: of individual spectra between two user-specified velocities,

349: similar to the print area command. }}

350: in CLASS.  We integrated over a range of

351: -9 to +8~km/s around the centroid velocity fit (in order to

352: include all hyperfine components), and take the error in the

353: integrated intensity to be $B \times \delta V / \sqrt{N}$,

354: where B is the spectrum's rms (baseline level), $\delta V$ 

355: is the velocity range (17~km~s$^{-1}$) and N the number of

356: spectral channels summed over (541).

357: 

358: To fit the single-line \cons(2-1) spectra, we used CLASS's

359: Gaussian fitting routine, and assumed a rest frequency of 219560.354~MHz 

360: \citep{Muller01}.  These spectra often had complex shapes not

361: well approximated by a single Gaussian.  \co traces less dense gas than \nh

362: and thus could be expected to more often trace multiple structures along

363: the line of sight.  The Gaussian profile, however, is a

364: simple approximation for the lines and provides a rough estimate

365: on relevant properties.  In cases where fitting a second Gaussian made a 

366: marked improvement to the fit, we did so (66 spectra).  We fit a third

367: component in only one case (a cross position) where three distinct and 

368: separate features were visible.  Figure~\ref{fig_sample_co} shows

369: three \co spectra as an example of some of the types of profiles

370: observed -- the first displays an obvious single Gaussian

371: profile, the second a clear double-Gaussian profile and the third

372: a more complicated structure which we fit with a single Gaussian.

373: 

374: Table~\ref{table_fits_co} shows the best-fit parameters found by CLASS

375: for the \co spectra - the centroid velocity, velocity dispersion 

376: (measured in FWHM units), integrated intensity, peak intensity, baseline (the

377: standard deviation where there is no line emission), and line rms 

378: (the standard deviation of the fitting residuals where there is

379: line emission).  Similarly to our procedure for the \nh lines, we 

380: take the error in the integrated intensity

381: (not included in the table) to be $B \times \delta V / \sqrt{N}$

382: where B is the baseline, $\delta V$ is twice the FWHM and N

383: the number of spectral channels summed over (each spectral channel

384: is 0.0267~km~s$^{-1}$). 

385: 

386: \subsection{Other Considerations -- Pointing Accuracy}

387: In the analysis below, we utilize only the measurements from the

388: central pointing and ignore the offset cross pointings that we have on

389: a subsample of the targets in order to treat the entire

390: sample consistently.  The observations

391: of cross-positions are useful in allowing us to determine how accurately

392: we determined the \nh core centres during target selection and how much error

393: or bias might be introduced to our results from using only a single

394: pointing.  We leave a full discussion of these issues for Appendix~C,

395: but note the result that we find little evidence our results are

396: affected by any errors or bias introduced by using a single pointing.

397: 

398: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

399: \section{NON-THERMAL MOTIONS WITHIN DENSE \nh CORES}

400: 

401: The one-dimensional thermal velocity dispersion expected for \nh is given by

402: \begin{equation}

403: \sigma_{T,n} = \sqrt{\frac{k_{B}~T}{\mu_{n}~m_{H}}}

404: \end{equation}

405: where $k_{B}$ is the Boltzmann constant, T is the temperature, $\mu_{n}$

406: is the molecular weight of \nh in atomic units (29) and $m_{H}$ the 

407: mass of a hydrogen atom.  We assume a temperature of 15~K, which lies

408: within the range of temperatures we derived from Bonnor-Ebert modelling

409: of SCUBA cores in the Perseus molecular cloud \citepalias{Kirk06}.

410: This corresponds to a sound speed of 0.065~km/s for \nhns.

411: The non-thermal component of the velocity dispersion is then given

412: by \begin{equation}

413: \sigma_{NT, n} = \sqrt{\sigma_{obs,n}^{2} - \sigma_{T,n}^{2} }

414: \end{equation}

415: where $\sigma_{obs,n}$ is the observed velocity dispersion.

416: {\hk The level of internal turbulence, the ratio of non-thermal velocity

417: dispersion to the mean thermal velocity dispersion of the gas, is then :}

418: \begin{equation}

419: f_{turb} = \frac{\sigma_{NT,n}}{c_{s}}

420: \end{equation}

421: where we take a mean molecular weight of 2.33, which yields a sound

422: speed of 0.23~km/s in the mean gas.  

423: 

424: We use the same procedure to calculate the level of non-thermal

425: motions observed in our \co data, with $\mu_{C}$ = 30 and a 

426: corresponding sound speed of 0.064~km/s.

427: 

428: The error in $f_{turb}$ is small for either line since the linewidths

429: are determined to better than 5$\%$ in most cases.

430: 

431: Recent ammonia observations (Rosolowsky et al, in prep), which include 

432: all of our targets, show

433: a spread in temperatures between 10 and 15~K, with the mean near 12~K.

434: The non-thermal motions we measure would only become $\sim$10\% larger

435: and the turbulent fraction $\sim$10 - 35\% larger with the adoption of 

436: the ammonia temperatures.  We maintain our use of

437: 15~K for consistency with \citetalias{Kirk06}.

438: 

439: Maps of dense cores show that they are surrounded by a less dense

440: envelope.  In targets where we detected an \nh signal, we expect

441: a surrounding envelope to also exist.  Due to chemical effects,

442: \co is expected to trace a larger scale than \nh even with its

443: smaller beamsize due to the higher frequency of the transition --

444: \co \citep[with critical density $\sim 10^{3}$~cm$^{-3}$, 

445: e.g.,][]{Ungerechts97} freezes out at densities of 10$^{5}$~cm$^{-3}$

446: where \nh is detectable \citep[e.g.,][]{Tafalla02}.  Therefore,

447: in a dense core, \co measures are weighted to the outer parts

448: while \nh measures are weighted to the denser inner parts.

449: Hence in targets where we detect both \nh and \cons, the \co can be

450: thought of as tracing the envelope of the dense \nh core.  

451: From our single pointing observations, we are unable to determine

452: whether this surrounding less dense gas is found distinctly around

453: each dense core (each core has a unique envelope) or on a larger

454: scale (several cores sharing a common envelope).  We therefore use

455: the term envelope broadly in our discussion of results.  In targets

456: where we only detect \cons, we do not have sufficient information

457: to determine if the emission originates from an envelope-like 

458: region.

459: 

460: {\hk Previous observations of dense cores and their surroundings have

461: shown that the dense gas (observed in \nh or NH$_{3}$) traces a `coherent

462: core' with a close-to-constant velocity dispersion of slightly above the

463: thermal value \citep{Benson89,Barranco98,Goodman98,Jijina99, Caselli02}.  

464: Additionally, the dense core appears kinematically distinct from

465: the surrounding less dense gas (traced by OH or C$^{18}$O) which

466: displays an increasing velocity dispersion with size 

467: \citep{Barranco98,Goodman98}.  We therefore expect that the velocity 

468: dispersion we measure in our \nh pointed observations represents

469: the value that would be present across the entire coherence length of

470: the core \citep[$\sim$0.1~pc for low-mass isolated cores in][]{Goodman98}.

471: Figure~\ref{fig_internal_turb_nh} which plots the distribution of the level

472: of internal turbulence ($f_{turb}$) measured in \nhns, shows that indeed

473: most of the dense cores have little non-thermal motion.}

474: 

475: Any dense cores which have evolved to the protostellar phase might be

476: expected to display a greater fraction of non-thermal motions -- 

477: either infall our outflow motion would be expected to broaden the

478: line width observed.  We analyze the subset of dense \nh cores which

479: are not associated with protostars from the Spitzer catalog (also

480: plotted in Figure~\ref{fig_internal_turb_nh}) and find that, as expected, 

481: this subset does tend to display even less turbulent motions.

482: The mean and standard deviation of the turbulent fraction is

483: 0.6$\pm$0.3 and 1.0$\pm$0.4 for the starless cores and protostellar 

484: cores respectively.

485: 

486: The turbulent fractions we find for starless cores and protostars 

487: are consistent with previous dense core surveys

488: \citep[e.g.][]{Benson89,Jijina99}.  While \citet{Jijina99} do observe

489: higher turbulent fractions in cores belonging to more massive and 

490: turbulent molecular clouds such as Orion, the turbulent fractions

491: we observe are consistent with the range \citet{Jijina99} find in

492: cores in molecular clouds with properties similar to Perseus.

493: 

494: The results are in contrast, however, to the

495: simulation of \citet{Klessen05} who find that their large-scale-driven

496: turbulence model only has $\sim$50\% of the cores with $f_{turb} \le 1$;

497: the small-scale-driven simulation has a much smaller fraction again.

498: The distribution of cores in the \citet{Klessen05} simulation are also

499: plotted in Figure~\ref{fig_internal_turb_nh}, displaying a significant

500: tail to the distribution of $f_{turb}$ of cores not found in our 

501: observations.

502: 

503: The turbulent fraction is also affected by association with young

504: stellar clusters.  \citet{Caselli95b} analyzed ammonia cores in the

505: Orion~B molecular cloud and found an inverse relationship between

506: core linewidth and distance to the nearest stellar cluster.  There

507: are three young star clusters near the Perseus molecular cloud --

508: in NGC1333, IC348, and the Perseus OB association

509: \citep[e.g.][]{Hatchell05}.  We note our observations are consistent

510: with a similar trend -- the most turbulent cores are found in

511: NGC1333 and IC348 (the Perseus OB association lies farther from the

512: cores we observed).

513: 

514: We can perform a similar turbulent fraction analysis with the 

515: \co observations (adapting

516: the velocity dispersion equations above).  The \co observations are 

517: sensitive to a lower density regime and so trace larger structures

518: which display a higher level of turbulent motion than the 

519: densest parts of the core displays (see Figure~\ref{fig_internal_turb_co}).  

520: This is also consistent with previous observations 

521: \citep[e.g.,][]{Benson89}.

522: The results from the \citet{Klessen05} model are also plotted for

523: reference.  Although the model is a much closer match to these observations,

524: the model `observations' were designed to match to a dense-gas tracer

525: such as \nhns.  The core boundaries used to define the `observable area' of

526: the cores in the simulation, the half-maximum column density contour

527: \citep{Klessen05} 

528: better match the extent of densities traced by \nh than by \cons.

529: It is interesting to note that our \co targets show much less variation

530: in their distribution of f$_{turb}$ between those which are and are not 

531: associated with a protostar than the dense \nh cores do.  The mean

532: and standard deviation are 1.2$\pm$0.6 and 1.8$\pm$0.7 for pointings

533: not associated and associated with protostars respectively.

534: This may indicate that at the early stages of protostellar evolution, 

535: protostars do not affect the bulk of their envelopes in 

536: a significant manner!  

537: 

538: %%%%%%%%%%%%%%%%%%

539: \section{CORE VERSUS ENVELOPE MOTIONS}

540: 

541: As discussed earlier, for the pointings which have both \nh and \co

542: detections, the \nh traces the dynamics of the dense core,

543: while the \co traces the dynamics of the surrounding less dense gas.

544: 

545: Previous studies using maps of \nh and \co have shown that cores

546: do not move ballistically within their envelopes (i.e. centroid

547: velocity differences between \nh and \co are smaller than the

548: \co linewidth) and most have subsonic core-to-envelope motions 

549: \citep[e.g.,][]{Walsh04, Walsh06}.  The \citet{Walsh04} survey examined

550: mostly single isolated cores, rather than those in clustered regions, while the 

551: \citet{Walsh06} survey spanned a clustered region of dense cores 

552: (NGC1333, which is included in our larger-area sample

553: although at lower resolution).  Unfortunately, the \citet{Walsh06} 

554: results are less certain since their \co beamsize was significantly 

555: larger than their \nh beamsize.

556: Thus our survey is ideal in providing a large statistical measure

557: of core-envelope motions within a clustered environment.  

558: 

559: We separately analyze the relative core to envelope motions of the starless 

560: and protostellar cores.  In starless cores, the relative motions

561: should be induced by the molecular cloud and the core formation

562: process.  In protostellar cores, the relative motions between

563: core and envelope could be complicated by outflows or processes which

564: decouple the core from its envelope.  Figure~\ref{fig_core_rel_envelope_1}

565: shows our results for the starless cores and protostars separately --

566: both have a high fraction of relative velocities which are less

567: than the sound speed.

568: 

569: 

570: \citet{Ayliffe06} has argued that the previous analyses of

571: \citet{Walsh04,Walsh06} biases results towards small velocity

572: differences due to the method of analysis.  In instances where

573: multiple CO velocities were found along the line of sight, 

574: the velocity component closest to the \nh core velocity was assumed

575: to be the one associated with the core.  Here, we demonstrate that

576: taking the closest CO velocity component is reasonable and does 

577: not introduce significant

578: bias.  Most of the cores in our observations had only a single

579: velocity component fit.  For these, we compared the difference in

580: centroid velocity between the \nh and \cons, as shown in 

581: Figure~\ref{fig_core_rel_envelope_1}.  This figure 

582: demonstrates that the vast majority of cores (nearly 90\%) have differences

583: less than the sound speed of the ambient medium (dotted lines), and

584: the remaining cores have differences which are not much larger.  

585: Note that some of the \nh cores

586: in this plot were fit to two velocity components; we considered

587: these two velocity components as separate entities (i.e., plotted

588: as two distinct cores).  The velocity differences found for the

589: two \nh velocity cores versus the surrounding \co tend to be

590: larger than for the other cores (since both \nh velocities are 

591: compared to the same \co velocity).  The CO linewidth for the two

592: velocity \nh cores also tends to be broader than average.

593: All of the cores have an error in the centroid velocity of

594: around several hundredths of a km~s$^{-1}$, indicating that the majority

595: of the differences in velocities observed are real.

596: 

597: Figure~\ref{fig_core_rel_envelope_2} shows the absolute difference

598: in \nh to \co centroid velocity in dense cores where

599: two \co velocity components were fit, again split into starless cores

600: and protostars.  The velocity differences here are ordered in terms of 

601: the largest-difference

602: component (squares), while the smallest-difference component 

603: is denoted by diamonds or asterisks.  Clearly the two CO velocity 

604: components are not correlated, as is expected since the second 

605: CO component merely lies along the same line of sight.  

606: The cores which happen to have a second \co velocity

607: component along the same line of sight should possess a similar distribution

608: of core-to-envelope relative velocities as 

609: along lines of sight where only a single \co velocity component

610: was observed -- i.e., velocity differences smaller than the sound speed in the 

611: vast majority of cases.  As can be seen from 

612: Figure~\ref{fig_core_rel_envelope_2}, in almost every case this implies the

613: closer CO velocity component is the only sensible one to associate

614: with the dense core; in the rare instance of ambiguity, the resultant

615: number of cores with each velocity difference will be little affected.

616: 

617: 

618: Assuming in all cases that the closest velocity \co component is the one

619: associated with the \nh profile peak, we find that in the majority of dense

620: cores, the core-to-envelope velocity tends to be smaller than thermal, 

621: (88\% and 83\% for the starless and protostellar cores respectively). 

622: In the \citet{Walsh04}

623: survey of mostly isolated cores, they find only one out of 35 cores

624: (or 3\%) with a core-to-envelope velocity exceeding the sound speed.

625: The \citet{Walsh06} survey of cores in NGC1333 found an rms

626: core-to-envelope velocity of 0.53~km/s; nearly half of their cores

627: have differences greater than the sound speed of the medium.  Our

628: survey includes the NGC1333 cores but finds much lower differences

629: (we have rms velocity differences of 0.16~km/s for the starless

630: cores and 0.18~km/s for the protostars when both the one- and two-

631: velocity component CO spectra are included).  This may indicate that 

632: the \citet{Walsh06} survey results were biased by the much larger

633: beamsize for \co (50\arcsec) than \nh (10\arcsec) which could have

634: sampled a large fraction of material not associated with the individual

635: dense core's envelope.

636: 

637: As discussed in \citet{Walsh04}, small relative motions between cores 

638: and envelopes could be interpreted

639: as an indication of quiescence on small scales, and as such would

640: appear to argue against a competitive accretion scenario for star

641: formation, where dense cores gain most of their mass by sweeping up

642: material as they move through the cloud.  Recent analysis of

643: simulations by \citet{Ayliffe06}, however, {\hk show} that the competitive 

644: accretion scenario is not necessarily incompatible with the observations.

645: \citet{Ayliffe06} analyzed

646: the competitive accretion simulations of \citet{Bate03} and demonstrated 

647: that the simulated observations also show core to envelope motions are

648: not ballistic.  There were, however, differences between the simulated

649: observations and the \citet{Walsh04} results at later times in the

650: simulation (such as the \nh linewidth becoming larger than the \co

651: linewidth), which were attributed to the clustered environment of the

652: simulation, versus the isolated cores observed.  Since our observations

653: probe cores forming in a clustered environment (and include protostars), 

654: we can make a stronger comparison with the \citet{Ayliffe06} results at

655: later times.  \citet{Ayliffe06} do find the majority of their

656: sources have velocity differences less than the sound speed at all

657: time steps, however, they have a more significant tail out to large

658: velocity differences (around 0.5~km/s or higher).  They find the 

659: dispersion in the velocity difference ranges from 0.25 to 0.27~km/s,

660: or 0.18 to 0.24~km/s after smoothing to the \citet{Walsh04} resolution.

661: It should also be 

662: noted that the \nh linewidth becomes equal to or larger than the

663: \co linewidth at 1.1 times the free-fall time and beyond, contradicting 

664: our observations (\S4) and that of many previous studies.

665: Thus while the simulation of competitive accretion analyzed by

666: \citet{Ayliffe06} has promise in reproducing core-to-envelope

667: dynamics it does not simultaneously reproduce all observations.

668: 

669: Our results show no indication that the core-to-envelope motions 

670: significantly change between the starless and protostellar stages of 

671: evolution.

672: 

673: On an even smaller scale, \citet{Jorgensen07} examined the location of

674: YSOs within SCUBA cores and showed that they lie within 15\arcsec of 

675: the SCUBA core centre.  These small separations imply that the YSOs

676: have motions smaller than the thermal velocity relative to the SCUBA

677: core they were born in.  The picture that emerges 

678: from the combination of these results is that the central source, core,

679: and envelope are quiescent.

680: 

681: %%%%%%%%%%%%%%%%%%

682: \section{CORE-TO-CORE MOTIONS}

683: We next examine how the \nh cores move with respect to each other to gain an

684: understanding of the dynamics on larger scales within the cloud.  

685: 

686: We can use our extinction map of the Perseus molecular cloud to

687: define the larger regions in which the dense \nh cores inhabit.

688: In \citetalias{Kirk06}, we identify large-scale structure in the extinction map 

689: which we will term `extinction regions' here to prevent confusion (the 

690: regions are referred to as `extinction super cores' in \citetalias{Kirk06}).

691: Starless cores within each extinction region should be coupled to the 

692: surrounding gas

693: in the region, and hence the motion of the starless cores should reflect 

694: the motion of the ambient material.  Protostellar cores may have become

695: decoupled from their parental material, and hence are a less reliable

696: tracer of the dynamics occuring in the region.

697: 

698: We analyze the motions within each extinction region and determine

699: whether the regions appear to have sufficient

700: velocity dispersion to provide support against gravity.  We adopt the

701: commonly used formulation of \begin{equation}

702: \sigma_{grav} = \sqrt{GM_{ext}/5R_{ext}}

703: \end{equation}

704: as the velocity dispersion required in 1-D to prevent collapse 

705: \citep[see for example][]{Bertoldi92}.  We estimate the total mass

706: and size of each region from the extinction data \citepalias{Kirk06};

707: {\hk these} data {\hk are} provided in Table~\ref{extinct_regions}.  

708: The above formula technically only applies to a uniform density sphere,

709: but different density structures and object shapes change 

710: the required velocity dispersion by factors of order

711: unity \citep{Walsh06,Bertoldi92}.  Regions which display 

712: $\sigma_{obs} = \sigma_{grav}$ are often said to be in approximate

713: virial equilibrium, although to have true virial equilibrium, the

714: `surface terms' of the virial equation must be included 

715: \citep[e.g.,][]{Dib06}. 

716: 

717: Motions providing support for the extinction region could

718: originate on either the small or large scale.  The former would

719: be measurable through internal core motions, while the latter 

720: through core-to-core motions.  In the case of the

721: densest material probed by \nhns, the internal core motions are of order

722: the thermal motions of a gas at $\sim$15~K.  If the extinction regions

723: were in virial equilibrium, they would require effective temperatures

724: of up to several hundred Kelvin to prevent gravitational collapse

725: \citepalias{Kirk06}; therefore, internal thermal motions cannot provide 

726: the bulk of the support required and hence most of the support must

727: originate in large scale motions.

728: We measure the dispersion in centroid velocities of the \nh starless

729: cores within each extinction region to determine the amount of support

730: that can be provided by large scale motions traced by the cores.  

731: We add the contribution of thermal motions (which has little effect

732: except for the smallest core-to-core velocity dispersions).  This total

733: support is plotted 

734: in Figure~\ref{fig_core_core} in terms of the ratio of observed velocity

735: dispersion to that which is required for virial equilibrium.

736: The horizontal axis plots the mass within each extinction region.

737: 

738: In the case of the material probed by \cons, small-scale motions could 

739: provide a larger contribution to overall support, since the internal 

740: velocity dispersions are often several times larger than the sound speed 

741: and are closer in magnitude to the point-to-point velocity differences.

742: In order to account for both of these contributions and also to

743: decrease potential errors from the difficulty in fitting each \co

744: spectrum,

745: we sum all of the spectra within an extinction region and fit

746: the sum with a single Gaussian thus measuring the total velocity

747: dispersion within each region.  We correct the thermal component

748: of the velocity dispersion to be that of the mean gas, rather than

749: \cons.  These results are shown in the plot as blue open diamonds.  

750: 

751: Table~\ref{extinct_regions} also summarizes the relevant information for

752: each extinction region -- the number of non-protostellar cores detected 

753: in \nh and \co in each region and the velocity dispersions

754: measured with both molecules.

755: 

756: Variations by a factor on the order of one could be expected between the

757: estimated and true velocity dispersion required for gravitational

758: support -- the extinction regions do not have a spherical geometry,

759: several extinction regions have a small number of cores to calculate the

760: velocity dispersion from, the cores do not span the entire extent 

761: of the extinction regions,

762: and the conversion between extinction and mass has some uncertainty.

763: It should also be noted that there is a velocity gradient across the

764: Perseus molecular cloud -- we leave a detailed analysis of the

765: core motions relative to the overall cloud gradient for a

766: future paper, and do not attempt to correct for it when calculating

767: the velocity dispersions used here.

768: With these considerations in mind, we find that the extinction

769: regions tend to display velocity dispersions lower than required for

770: `virial equilibrium', with the starless cores possessing lower 

771: dispersions in \nh than in \cons, but the measurements do not rule

772: out `virial equilibrium'.  

773: 

774: %%%%%%%%%%%%%%%%%

775: \section{ENVIRONMENTAL EFFECT ON DENSE CORES}

776: {\hk Most (84\%) of the SCUBA cores have associated \nhns.  For these 

777: cores, we can examine whether the SCUBA core properties have an 

778: effect on their internal dynamics.  Since \nh requires 

779: $\sim 10^{5}$ years to form \citep{Aikawa03}, the SCUBA cores must be

780: at least this old, and hence are at an advanced stage of

781: evolution, consistent with the results of \citet{Jorgensen07}. }

782: 

783: \subsection{Concentration}

784: The `peakiness' or central concentration of a core gives an 

785: indication of the importance of self-gravity of the core.

786: The concentration is defined in terms of observable measures

787: in equation (A1).

788: Concentration can be thought of as an approximate proxy for evolutionary

789: state with high concentration objects being more evolved 

790: \citep{Walawend05,Walawend06,Johnstone06,Jorgensen07} -- in the framework of 

791: a Bonnor Ebert sphere model, any object with concentration above

792: 0.72 is unstable to gravitational collapse, furthermore, heating from

793: a central source also leads to an increase in concentration. 

794: Figure~\ref{fig_internal_turb_concs}

795: shows the variation in observed core velocity dispersion with SCUBA 

796: concentration

797: for both protostellar and starless cores.  The mean concentration and 

798: velocity dispersion

799: are lower in the starless cores than the protostars,

800: with protostars of higher velocity dispersion also possessing high

801: concentrations.

802: The mean and standard deviation of the concentration is C = 0.4$\pm$0.1

803: and C = 0.6$\pm$0.2 for the starless and protostellar cores respectively.

804: The velocity dispersion observed for cores associated with a SCUBA source

805: is 0.20$\pm$0.08~km/s for the starless cores and 0.25$\pm$0.09~km/s for the 

806: protostars.

807: The scatter in the velocity dispersion of the starless cores is mostly

808: due to two starless cores with unusually high internal 

809: turbulence levels ($>$ 1.5); excluding these two cores, the velocity

810: dispersion becomes 0.18$\pm$0.04~km/s.  These

811: two starless cores are in NGC1333 where the region is highly clustered,

812: making it more difficult to determine accurate core properties as

813: well as determine if there is an associated protostar.

814: The velocity dispersion for those cores which are not associated

815: with a SCUBA source (primarily targets selected from the Palomar

816: plates) tend to be even lower -- the mean and standard

817: deviation is 0.14$\pm$0.04~km/s.

818: 

819: \subsection{Total Flux}

820: We next examine the relationship between internal turbulence

821: level with the total flux of the SCUBA core.

822: Figure~\ref{fig_internal_turb_flux} shows SCUBA core total flux versus

823: the velocity dispersion for both the starless and protostellar 

824: cores.  No trend is apparent for the starless cores (diamonds), while

825: there appears to be a weak trend of higher flux corresponding to higher 

826: velocity dispersion in the protostars (asterisks).  If we split the protostars

827: into those with fluxes greater than 5~Jy and less than 5~Jy,

828: we find the mean and standard deviation are 0.31$\pm$0.10~km/s and 

829: 0.23$\pm$0.07~km/s respectively.  The {\hk symbols intersected by

830: crosses} show the mean and standard deviation for the

831: protostars and starless cores, 

832: indicating that the protostars tend to have higher flux than the 

833: starless cores.  This could be the result of slightly higher central 

834: temperatures in the protostellar cores.

835: 

836: If we assume a constant temperature of 15~K, a dust opacity of 

837: 0.02~g$^{-1}$cm$^{2}$ at 850~$\mu m$ and a distance to the

838: Perseus molecular cloud of 250~pc \citep[e.g.,][]{Cernis93}, we can convert

839: the observed SCUBA flux into mass as 1~Jy = 0.48~M$_{\odot}$ 

840: \citepalias[see][]{Kirk06}.  Note that due to the non-negligible 

841: uncertainties in all of 

842: the above quantities, the mass is only accurate to a factor of 

843: roughly 6.  Even with the large uncertainty, 

844: we can use the mass and radius measured for each SCUBA core to

845: estimate the internal velocity dispersion required to provide

846: support against gravity, which we again take to be 

847: $\sigma_{grav} = \sqrt{GM_{C}/5R_{C}}$ (c.f. eq. [4]).

848: 

849: The velocity dispersion of the mean gas can be calculated by correcting

850: for a thermal component with a molecular weight of 2.33 rather than

851: 29 for \nhns.  We can then compare the total gas velocity dispersion to

852: that predicted for virial equilibrium (from the SCUBA observations).

853: Figure~\ref{fig_internal_turb_mass} shows the square of the ratio of the 

854: observed velocity dispersion to the virial velocity versus SCUBA flux 

855: for the starless cores and protostars. `Virial equilibrium' would

856: occur for a ratio of 1 (dotted line).  All of the \nh cores have

857: higher velocity dispersions than predicted by the virial equation,

858: with those at small SCUBA fluxes displaying the largest difference.

859: The cores farthest

860: from `virial equilibrium' would require the SCUBA mass to be underestimated by

861: a factor of ten or more if they were truly in virial equilibrium, far

862: larger than our uncertainties allow.  Many of the cores would be far from

863: virial equilibrium even if their observed velocity dispersion were purely

864: thermal -- the dashed and dot-dashed

865: lines show the relationships for a 15~K thermal core velocity dispersion 

866: and assuming core radii of 10\arcsec\ and 60\arcsec\ respectively 

867: (bounding the observed range of SCUBA core radii).

868: 

869: The above analysis ignores the contribution of external pressure

870: in the virial equilbrium calculation which for sub-Jeans mass objects

871: keeps the internal motions thermal even though gravity alone does not

872: require internal motions of this magnitude.  {\hk In \citetalias{Kirk06}, we} 

873: find that

874: the SCUBA cores in Perseus are well fit by Bonnor-Ebert spheres with

875: external pressures in the range of 5.5~$\le$~log$_{10}$P$_{ext}$/k$_{B}$

876: $\le$~6.0.

877: For a `typical' core of roughly one solar mass and 50\arcsec\ in extent,

878: when the external pressure is included in calculating virial equilibrium,

879: the square of the virial velocity rises by a factor of approximately 1.3 

880: to 2 of what it was without considering the external pressure, which 

881: would make most of

882: the higher flux cores in approximate virial equilibrium.  

883: Similarly, a critical BE sphere, which has

884: R$_{crit}$=0.41$\frac{GM}{c_{s}^{2}}$ \citep{Hartmann98}, 

885: requires $\sqrt{5\times0.41}$, or roughly 1.4, times the velocity

886: dispersion one would naively assume. 

887: Note that observing a line width and converting to mass using the virial

888: equation without accounting for surface pressure will overestimate the

889: enclosed mass of an equilibrium core.  See also \citet{Dib06}.

890: 

891: Our results thus show that while we observe velocity dispersions

892: that are several times larger than what is predicted by the

893: traditional `virial equilibrium' measures, when the external

894: pressure on the dense cores from the ambient cloud is accounted

895: for, the agreement is reasonable for the higher flux cores.

896: This is in contrast with some previous observations which

897: tend to find velocity dispersions which are consistent

898: with `virial equilibrium' without accounting for any

899: external pressure -- for example, the low mass dense core

900: survey of \citet{Caselli02}.  Some previous studies, however, 

901: have shown that external pressure is required for virial equilibrium, 

902: e.g., in the Horsehead nebula \citep{Wardthomp06}.

903: 

904: The turbulent simulations of \citet{Klessen05} predict a 

905: relationship between virial and observed mass -- 

906: their large-scale-driven simulation (which more closely matches our

907: other observations) shows that the starless cores have

908: virial masses which are greater by up to a factor of thirty 

909: than the actual mass.  Unlike our observations, however,

910: the simulation shows that protostars have virial masses which

911: are several times less than the actual mass.  

912: \citet{Klessen05} points out that the virial mass estimates

913: for protostars are underestimated due to the lack of

914: velocity resolution of the gas in the central sink cell, but that

915: this will have a small effect on the measured velocity dispersion

916: since a small fraction of the core mass is contained within the

917: sink cell.

918: 

919: \subsection{Variation of Line Intensity}

920: 

921: \nh and \co observations can also serve as a probe of the chemistry

922: of the dense cores.  \nh is only able to form

923: in significant amounts after \co freezeout has occurred, as the two

924: molecules form via competing reactions.  While \nh is observed to be a good

925: dense gas tracer for densities of $10^{5} - 10^{6}$~cm$^{-3}$

926: \citep{Tafalla02}, it may freeze out onto dust grains at densities 

927: above this \citep{Crapsi05}.  \co on the other hand, should be depleted

928: at high densities.  

929: At later stages of evolution once a central protostar has formed,

930: the situation is expected to reverse, with the central region heating, 

931: causing the liberation of CO and destruction of \nhns.

932: Our SCUBA observations allow us to estimate the (column) density of 

933: the dense cores independently of our IRAM observations which may 

934: be affected by chemistry.  Using the

935: same flux to mass conversion factors discussed above, we can

936: convert the SCUBA flux into a column density --

937: 1~Jy~beam$^{-1}$ corresponds to 0.24~g~cm$^{-2}$ or $\sim10^{22}$~cm$^{-2}$.  

938: Making the further assumption

939: that the cores are roughly spherical and have a diameter of $\sim$~50\arcsec\

940: in the plane of the sky, this corresponds to a density of 

941: $\sim 10^{6}$~cm$^{-3}$.

942: 

943: Figure~\ref{fig_intens_vs_submm} shows the total SCUBA flux within the

944: IRAM beam for each observation versus the integrated intensity measured 

945: in both \nh and \cons.  The \nh integrated intensity shows some correlation

946: with the total SCUBA flux for both the starless and protostellar cores. 

947: The \co integrated intensity possibly shows a very weak correlation with

948: SCUBA flux below $\sim$1Jy and no correlation above this.  This is consistent

949: with denser cores being dominated by central freeze out 

950: (i.e., even with increasing column density, the 

951: \co integrated intensity remains constant). 

952: 

953: We also examine the

954: ratio between the \co and \nh integrated intensities -- a low ratio, for 

955: example, would be indicative of freezeout.

956: Figure~\ref{fig_intens_ratio_vs_submm} shows the total SCUBA flux

957: observed in each IRAM beam versus the ratio of integrated intensity measured

958: in \co and \nhns.  This figure shows that high \co to \nh ratios mostly

959: occur for starless cores, and only at smaller SCUBA fluxes, i.e., where 

960: the density is lowest and hence there is little to no freeze out.  

961: All high flux SCUBA cores have low \co to \nh ratios.  Due to

962: the large relative error in the \nh integrated intensity for these

963: cores, the error in the ratio is often greater than 100\%.

964: Note that following

965: the results of \S5, in the few cases where two \co components were

966: associated with an \nh dense core, the integrated intensity of the

967: component with the closest velocity is plotted.

968: 

969: We can also calculate the \nh column density from the integrated intensity,

970: assuming an excitation temperature of 15~K and correcting for the optical

971: depth.  We use eq. (10) of \citet{Shirley05}, and find a column density of

972: \begin{equation}

973: N_{N_{2}H^{+}} = 1.47\times10^{7} \frac{\langle \tau \rangle}

974: {1-e^{-\langle \tau \rangle}} 

975: \int T_{A}^{*} dV \times \frac{F_{eff}}{B_{eff}}\  cm^{-2}

976: \end{equation}

977: where $\langle \tau \rangle$ is the mean optical depth of the hyperfine

978: transitions (CLASS's hfs fitting routine fits 

979: $\tau_{tot} = 7 \langle \tau \rangle$), $\int T_{A}^{*} dV$ is the

980: integrated intensity in K~m~s$^{-1}$, and $F_{eff} = 0.95$ and 

981: $B_{eff} = 0.77$ are

982: beam efficiency parameters available from the IRAM 30m website.

983: Our minimum observable column density is $\sim 10^{11}$~cm$^{-2}$.

984: 

985: Figure~\ref{fig_coldens} shows the \nh column density derived versus

986: the total column density derived from the total SCUBA flux measured in the

987: IRAM beam.  The relative error in the total column densities is

988: $\sim$30\% (the calibration error of SCUBA data), with an absolute error

989: close to a factor of six due to uncertainties in constants used to convert

990: flux to mass.  The errors in the \nh column density vary substantially,

991: mostly due to errors in the optical depth determined; the median error is

992: 30\%.  The noise in the SCUBA map is $\sim$10~mJy~beam$^{-1}$, which 

993: translates to

994: a minimum observable total column density of $\sim 10^{21}$~cm$^{-2}$.

995: 

996: Overplotted on Figure~\ref{fig_coldens} are lines of constant \nh abundance.

997: The cores lie between approximately N$_{N_{2}H^{+}}$ / N$_{H_{2}} = 10^{-9}$ and

998: 10$^{-10}$, 

999: consistent with what has been more accurately derived from detailed mapping

1000: and analysis of single cores \citep[e.g.,][]{Shirley05,Tafalla04}.

1001: 

1002: %%%%%%%%%%%%%%%%%%

1003: \section{CONCLUSIONS}

1004: We present results from a survey of \nhns(1-0) and \cons(2-1) of 157 dense 

1005: core candidates in the Perseus molecular cloud.  We detect \nh in

1006: 62\% of our targets, and 84\% of our SCUBA-selected targets.  

1007: \nh is a `late-time' {\hk molecular ion} which does not become abundant until

1008: $\sim10^{5}$~years \citep{Aikawa03}.  Since we detect \nh in the vast majority

1009: of SCUBA cores, this argues that objects which attain sufficient

1010: density to be detectable with SCUBA are not short-lived, transient

1011: objects.  This is in agreement with the findings of \citet{Jorgensen07}

1012: who argue that starless SCUBA cores have roughly equal lifetimes to

1013: that of deeply embedded protostars, which is on the order of $10^{5}$ years

1014: \citep{Wardthomp07}. 

1015: 

1016: We differentiated between starless cores and protostars

1017: on the basis of Spitzer data \citep{Jorgensen07}.  In \nhns, the starless

1018: cores have linewidths which are dominated by thermal broadening, while

1019: the protostars have slightly larger linewidths, consistent

1020: with many previous surveys including \citet{Benson89} and \citet{Jijina99}.

1021: We find fewer \nh cores dominated by non-thermal motions than 

1022: predicted by the turbulent simulations of \citet{Klessen05}.

1023: For the starless cores, the mean ratio of non-thermal to thermal motions

1024: ($f_{turb} \sim 0.6$) implies the ratio of non-thermal to thermal

1025: pressure, $f_{turb}^{2}$, is less than 40\%.  Naively, this runs

1026: counter to turbulent models where cores as well as transient density

1027: peaks form at the convergence of supersonic flows.  Simulations must

1028: therefore demonstrate that turbulent pressure does not dominate

1029: in the high density regime probed by \nhns.

1030: 

1031: The \co observations, sensitive to lower density material, reveal

1032: much more non-thermal motions as previous surveys have also shown.  There is 

1033: less difference in non-thermal motions between

1034: those targets associated with protostars and those which {\hk are not} associated

1035: with protostars

1036: which may imply that protostars have little effect on the dynamics of bulk

1037: of their envelopes at the earlier stages of evolution.

1038: 

1039: We find the relative motions of the dense \nh cores and their envelopes 

1040: (measured in \cons) tends to be less than thermal in the majority of cases,

1041: confirming and strengthening the results of \citet{Walsh04,Walsh06}

1042: for clustered star forming environments.  

1043: 

1044: Within large scale structure, defined through 2MASS extinction observations, 

1045: the core-to-core motions of starless

1046: cores are not sufficient to provide support against gravity, however,

1047: the sparse sampling of each extinction region leads to large errors

1048: associated with the core-to-core velocity dispersions we measure.

1049: The total velocity dispersion tends to be smaller when measured in

1050: \nh than \co due to the smaller linewidths seen in \nhns.

1051: 

1052: The \nh cores which have an associated submillimetre source detected with 

1053: SCUBA have internal motions several times larger than is required 

1054: to provide support against gravity.

1055: Inclusion of external pressure shows the cores to be in approximate

1056: virial equilibrium.  The protostars tended to have higher SCUBA concentrations,

1057: and total fluxes.

1058: High ratios of \co to \nh integrated intensity, possibly indicating

1059: chemically young gas, were found for some cores which had a low flux

1060: measured with SCUBA.  At higher SCUBA fluxes,

1061: only low ratios of \co to \nh integrated intensity were observed.

1062: Column densities derived for \nh were consistent with abundance ratios

1063: between 10$^{-9}$ and 10$^{-10}$, in agreement with what has been 

1064: previously derived for cores with more accurate observations.

1065: 

1066: Our survey utilized single pointings on most of our dense core

1067: candidates, rather than using the traditional route of mapping.  We

1068: show that the lack of a full map around each source has a minimal

1069: effect on our dynamical analysis.  High resolution maps are, however, quite

1070: helpful in disentagling the motions in complex regions such as

1071: NGC1333 where multiple objects along the line of sight could otherwise

1072: lead to confusion in interpretation of results.  Our method is

1073: an efficient and effective way to survey the dynamics of dense cores

1074: over the full extent of a molecular cloud.  In the future, this

1075: method can be applied to other molecular clouds in order to

1076: determine whether the dynamical properties of the cores observed

1077: in the Perseus molecular cloud are universal or are dependent

1078: on the cloud environment.

1079: %%%%%%%%%%%%%%%%%%%%%%%%%%

1080: 

1081: \acknowledgements

1082: We thank the IRAM 30~m staff for their hospitality and support during our 

1083: observations.

1084: 

1085: The Second Palomar Observatory Sky Survey (POSS-II) was made by

1086: the California Institute of Technology with funds from the National

1087: Science Foundation, the National Geographic Society, the Sloan

1088: Foundation, the Samuel Oschin Foundation, and the Eastman Kodak

1089: Corporation.

1090: 

1091: HK\footnote{Guest User, Canadian Astronomy Data Centre, which is 

1092: operated by the Herzberg Institute of Astrophysics, National Research 

1093: Council of Canada} 

1094: is supported by a Natural Sciences and Engineering Research Council

1095: of Canada CGS Award and a

1096: National Research Council of Canada GSSSP Award.  DJ is supported 

1097: by a Natural Sciences and Engineering Research Council of Canada 

1098: grant.

1099: 

1100: HK would like to acknowledge valuable discussions with people at the

1101: CfA -- other members of the COMPLETE collaboration, in particular

1102: Alyssa Goodman, Jens Kauffman, and Erik Rosolowsky, as well as 

1103: Phil Myers and Charles Lada.  Additionally, HK thanks Matthew Bate 

1104: for a sending a preprint of his work.

1105: %%%%%%%%%%%%%%%%%%%%%

1106: 

1107: \appendix

1108: 

1109: \section{SCUBA OBSERVATIONS}

1110: In this paper, we utilize newly created 850~$\mu$m SCUBA maps with a finer

1111: sampling size of 3\arcsec\ to better determine SCUBA source properties.

1112: We used our full 6\arcsec\ sampled map of \citetalias{Kirk06} to define

1113: regions in which to create 3\arcsec\ sampled maps.  As in \citetalias{Kirk06},

1114: we combined all of the scan- and jiggle- map data in the JCMT 

1115: archive\footnote{Based on observations obtained with the James Clerk Maxwell 

1116: Telescope, which is operated by the Joint Astronomy Centre in Hilo, Hawaii 

1117: on behalf of the parent organizations PPARC in the United Kingdom, the 

1118: National Research Council of Canada and The Netherlands Organization for 

1119: Scientific Research.} 

1120: and used the same reduction procedure as

1121: in \citetalias{Kirk06}, making modifications only for the smaller sampling size.

1122: We first use the normal SCUBA software \citep{Holland99} to flat-field and

1123: atmospheric-extinction correct the raw data.  We then used the

1124: matrix inversion technique of \citet{Johnstone00} to produce the images.

1125: The matrix inversion techniqe has been shown to produce better

1126: images than the standard procedure used at the JCMT, particularly

1127: when combining data of different qualities \citep{Johnstone00}, as is the

1128: case for the archival data used here.  In order to correct for atmospheric

1129: fluctuations and other effects, SCUBA data is in the form of a series of 

1130: difference measures (chops).  Any image-reconstruction technique is 

1131: thus insensitive to real structures which have sizes several times

1132: larger than the chop throw \citep{Johnstone00}.

1133: We remove this structure by subtracting a large-scale

1134: smoothed version of the map from the original (we smooth with a Gaussian

1135: of $\sigma = $90\arcsec).  In order to prevent the introduction of 

1136: negative `bowls' around bright sources (and similarly diffuse

1137: positive regions around deep compact `holes'), we first create a

1138: map where all values outside of $\pm 0.1$~Jy~beam$^{-1}$ per pixel 

1139: were replaced with those

1140: values before smoothing to create the large-scale smoothed map 

1141: (0.1~Jy~beam$^{-1}$ corresponds to roughly five times the rms value).  We 

1142: also smoothed pixel-to-pixel noise using a Gaussian with $\sigma = 3$\arcsec. 

1143: Figure~\ref{fig_submm_ngc1333} shows our map of {\hk B1} as an example

1144: of the 3\arcsec\ SCUBA maps.

1145: 

1146: To identify SCUBA cores in the maps, we utilized the object-identifying

1147: algorithm of \citet{Williams94}, `Clumpfind 2D'.  In this algorithm, objects

1148: are identified as peaks at 2~`$\sigma_{C}$' intervals and extended until 

1149: they either encounter another object or the lowest allowed `$\sigma_{C}$'

1150: level.  Normally, $\sigma_{C}$ is the noise level in the map, 

1151: however, in order to have a consistent core-identification threshold

1152: in all of the 3\arcsec\ mapped regions, we used $\sigma_{C} =$ 

1153: 0.03~Jy~beam$^{-1}$

1154: per pixel, which corresponds to approximately the same level in which we 

1155: identified SCUBA cores in our previous work \citetalias{Kirk06}.  Accurate 

1156: noise levels for each 3\arcsec\ sampled map were difficult

1157: to determine in some cases due to the small map sizes.  

1158: 

1159: In addition, several regions had no cores identified

1160: but displayed hints of structures with peaks below the object-identification 

1161: theshold of 5$\sigma_{C}$.  In order to put some constraints / 

1162: upper limits on the submillimetre properties of potential cores in these

1163: regions, we ran clumpfind to a lower identification threshold 

1164: ($\sigma_{C}$ = 0.01~Jy~beam$^{-1}$ per pixel) in these regions only.  

1165: It should 

1166: be noted that properties derived for these objects are not as reliable 

1167: as the originally identified cores.  We term these objects as having 

1168: `less secure fits' throughout the paper and do not include these in our 

1169: quantitative analysis.  Table~\ref{scuba_appendix} below denotes the

1170: properties of the SCUBA cores identified.

1171: 

1172: Finally, as discussed in Appendix~C, SCUBA observations in the region of NGC1333

1173: appear to have a shift of 6\arcsec\ in RA relative to data at other

1174: wavelengths, apparently due to an unusually large pointing error

1175: at the JCMT.  Here, we apply a global shift of 6\arcsec\ to the

1176: NGC1333 observations to compensate for this.

1177: 

1178: \subsection{Comparison to Previous Results}

1179: While Clumpfind does a reasonable job of identifying structures in two 

1180: dimensional maps where the filling factor is low, the cores identified in

1181: the 3\arcsec\ sampled maps used here are different from the set we 

1182: identified previously in the 6\arcsec\ sampled map of \citetalias{Kirk06}, 

1183: even though the same data {\hk are}

1184: used and the reduction procedure is almost identical, albeit with

1185: a different smoothing scale.  This is

1186: because Clumpfind relies on contours for determining object edges, so that

1187: slight variations in flux per pixel can change the size of core boundaries, 

1188: which then affects the measured size and total flux, although the peak

1189: flux would be unchanged.  In clustered regions, the slight variation in 

1190: flux per pixel can

1191: also change where or when cores are either separated from or merged with 

1192: close neighbours.  Clumpfind identifies distinct clumps where two regions are

1193: isolated at a given search contour (every 2$\sigma_{C}$).

1194: For example, a peak at 6.9$\sigma_{C}$ surrounded by pixels at 6.1$\sigma_{C}$ 

1195: and near a peak at 11$\sigma_{C}$ would be identified as a single object 

1196: (at the 7 and 5$\sigma_{C}$ contours, all of the flux is connected).  If

1197: the 6.9$\sigma_{C}$ peak were instead a peak at 7.1$\sigma_{C}$, 

1198: it would be identified as a separate object from the 11$\sigma_{C}$ peak 

1199: (the flux is in two unconnected regions at the 7$\sigma_{C}$ 

1200: contour).  Thus individual core properties are difficult to compare

1201: between maps reduced under even slightly different schemes or

1202: resolutions.  We do not show a comparison

1203: of the cores identified here with those identified in the 6\arcsec\ sampled

1204: map of \citetalias{Kirk06}, but note that given variations in

1205: core boundaries and potential merging of cores, the list of cores we 

1206: identify in the 3\arcsec\ sampled map spans the cores identified in the 

1207: 6\arcsec\ sampled map and also includes an additional four cores

1208: (\#4, 28, 72, 73).

1209: 

1210: \subsection{Core Properties}

1211: The properties of the cores identified in the 3\arcsec\ sampled map are

1212: shown in Table~\ref{scuba_appendix}.  The core radius, peak flux, and total flux

1213: are found with Clumpfind.  We also calculate the concentration of

1214: each core -- previous work has shown this to be an indicator

1215: of the evolutionary state of the core, with higher concentrations

1216: corresponding to later stages of evolution

1217: \citep{Walawend05,Walawend06,Johnstone06}.

1218: Following \citet{Johnstone01}, the concentration can be calculated

1219: from observational quantities as: 

1220: \begin{equation}

1221: C = 1 - \frac{1.13 B^{2}S_{850}}{(\pi R^{2}_{obs})f_{0}}

1222: \end{equation}

1223: where B is the beamsize, $S_{850}$ the total flux, $R_{obs}$ the

1224: radius, and $f_{0}$ the peak flux.  As in \citetalias{Kirk06}, we can

1225: also model the cores as Bonnor Ebert (BE) spheres -- spherically symmetric

1226: isothermal objects where thermal pressure balances gravity and an 

1227: external pressure.  The best fit BE sphere model properties are also

1228: included in Table~\ref{scuba_appendix}.

1229: 

1230: %%%%%%%%%

1231: \section{TWO-COMPONENT \nh CORES}

1232: 

1233: In this section we discuss the \nh spectra which we found required two

1234: velocity components for a good model fit.

1235: We treated these two velocity components as originating from independent

1236: objects along the line of sight, rather that a single core whose spectrum

1237: shows self-absorption.  The optimal method for distinguishing between

1238: self-absorption and two distinct cores would be to observe the region

1239: with an optically thinner tracer.  Since this is not available to us, we

1240: instead examine the data we do have for these cores (e.g. spectra at

1241: cross positions, SCUBA observations, and other \nh survey data) and

1242: discuss how well they support our interpretation. 

1243: 

1244: Most of the cores for which we fit to two components lie in regions where

1245: complex motions are seen on smaller scales, such as NGC1333, where it

1246: is not unsurprising to find two cores along the line of sight.  

1247: In instances where we found a common velocity component between two

1248: close pointings, we excluded this common velocity component from

1249: the second source in our subsequent analysis in order

1250: to avoid counting the same core twice.

1251: 

1252: In NGC1333, we found six cores with two

1253: velocity components -- \#99, 103, 106, 107, 111, and 118. 

1254: All of these except \#99 fall within the survey region

1255: of \citet{Walsh06} which in every instance identifies a distinct

1256: object at each velocity which we fit to our data.  Source \#99, which falls

1257: outside of the \citet{Walsh06} survey region, was observed at cross positions 

1258: (see Figure~\ref{fig_ps_s36}).  The two components show 

1259: varying relative intensities across the

1260: five positions mapped, supporting the hypothesis of two distinct cores.

1261: 

1262: 

1263: In IC348, we fit two cores with two velocity components, \#22 and 27.  

1264: While there is no high resolution \nh map of IC348, \citet{Tafalla06}

1265: provide a 50\arcsec\ resolution map.  Our core \#22 corresponds to

1266: \citet{Tafalla06}'s source C in IC348-SW1 which they show appears to

1267: be related to outflow structure seen in CO.  \citet{Tafalla06}

1268: identify two velocity components in \nh which correspond to the velocities

1269: we found.  Our core \#27 corresponds to source A in IC348-SW1

1270: which \citet{Tafalla06} associate with a single broad ($\sim$1.6~km/s)

1271: velocity component.  Our observations (Figure~\ref{IC348}) clearly 

1272: show two distinct

1273: velocity components, perhaps resolvable due to the smaller beamsize of

1274: our observations. 

1275: 

1276: 

1277: In L1544, one core (\#136) was fit with two velocity components.

1278: Cross positions were also observed, and where the S/N is high enough, 

1279: the 

1280: two components are clearly quite separate (see Figure~\ref{fig_ps_s41}).

1281: 

1282: North of B1, one core (\#76) was fit with two velocity components at

1283: one cross position only.  This observation has low S/N and is designated as 

1284: being of poor quality, and therefore was not used in any of the analysis.

1285: 

1286: The core with the poorest two velocity component fit is \#148 in L1448.

1287: In this instance, the model does not well fit the data in several places

1288: (see Figure~\ref{L1448}), making the interpretation of two distinct cores 

1289: uncertain.  The SCUBA map suggests the core is isolated -- the nearest

1290: SCUBA core is roughly an arcminute away.  Without

1291: additional observations, no firm interpretation

1292: can be made.  

1293: 

1294: %%%%%%%%%%%%%%%%%%%%%%%%%%

1295: \section{EFFECT OF UTILIZING SINGLE POINTINGS}

1296: For the analysis in the paper, we utilized only the data from

1297: the central pointing on each target in order to have a consistent

1298: dataset.  Since we have four-point cross data around some of our

1299: targets, we can ask two additional questions -- how accurate were

1300: the target centre positions that we used and does our utilization

1301: of only the central point bias or change any of our results.  Here we

1302: address these questions in turn.

1303: 

1304: \subsection{Accuracy of Dense Core Pointings}

1305: We first note one occurence that led to small offsets between some

1306: positions in the SCUBA catalog we used for determining our target 

1307: positions and the catalog of Appendix~A.  

1308: After the publication of \citetalias{Kirk06}, we discovered an offset 

1309: of $\sim$6\arcsec\ in RA 

1310: in the region of NGC1333 in the SCUBA data compared with data 

1311: at other wavelengths.  This is intrinsic to the data (not caused by

1312: an analysis error) and is apparently caused by an unusually large 

1313: pointing error with SCUBA.  This offset is further discussed in 

1314: \citet{DiFrancesco07}.  This offset was not known at the time of

1315: our IRAM observations, hence was not accounted for when the 

1316: SCUBA target list was created.

1317: This 6\arcsec\ offset is much smaller 

1318: than the IRAM beamsize (25\arcsec) and hence should not have a large 

1319: effect on the results, but is noted here for completeness.

1320: 

1321: We expect the \nh cores to have extents of order one or two 

1322: IRAM beams (25\arcsec) in \nhns, the typical size of the

1323: SCUBA cores.  SCUBA is sensitive to a similar range of densities, 

1324: $\sim 10^{4}-10^{6}$~cm$^{-3}$ using the approximation

1325: in \S7.  If we chose our target positions well, we expect the central 

1326: position to show more signal than offset positions, although the cores

1327: should extend past the central pointing.

1328: 

1329: Figure~\ref{fig_cross_area} shows the fractional

1330: difference between the integrated intensity at the central position

1331: and largest value at an offset position (pluses) as well as the

1332: difference with the average value of all offset positions (squares).

1333: The vertical lines indicate the error in the difference measure for

1334: the maximum difference.  Nearly half of the cores plotted

1335: have highest integrated intensities at the central pointing,

1336: while another third have integrated intensities which are slightly

1337: larger at a single cross position.  Only about one fifth of the cores 

1338: show significantly larger integrated intensities at a single offset position.

1339: An additional thirteen positions where offset observations were made are not

1340: plotted due to a lack of a good central detection.  Of these, ten also

1341: had no detections at any offset positions.

1342: 

1343: It should be noted that we utilized two different set of criteria to

1344: determine which targets to map crosses around.

1345: Approximately half (thirty of sixty-two) of the cross targets

1346: had strong central detections -- we observed at offsets to search for  

1347: extended structure.  The remaining half of the targets

1348: were chosen based on a very

1349: weak or non-existent central detection in order to search for a 

1350: stronger signal nearby (2MASS or Palomar plate - selected

1351: targets).  The first of these cross-map criteria biases

1352: towards targets where the centre was well chosen, while the second 

1353: biases towards poorly chosen centres.

1354: The former set of cross maps were mostly based on SCUBA cores

1355: where we had a high detection probability and it was easiest to

1356: determine precise core centres.  Nearly all of the SCUBA-selected 

1357: cores with crosses are consistent with having 

1358: their strongest integrated intensity at the central position.

1359: The latter-chosen cross targets account for the largest integrated

1360: intensity differences at

1361: cross positions as well as the targets with insufficient signal to noise to

1362: obtain a fit and thus be included in the plot.

1363: Not surprisingly, our position selection was less accurate for these 

1364: latter targets -- many of the detections are not consistent with

1365: the central position having the highest integrated intensity.  

1366: Since these latter crossed observations

1367: are biased to the worst-determined centre positions, we expect

1368: that our overall determination of centres was more successful.

1369: 

1370: We can therefore conclude that the pointings were quite accurate for the

1371: SCUBA-selected targets and reasonably accurate for the targets selected 

1372: using the

1373: other methods, especially given the difficulties

1374: in determining precise centres in the latter case.  As we shall see in the

1375: following section, the question of accuracy of the centre positions does

1376: not play an important role in the kinematical results presented earlier --

1377: these change very little between centre and cross positions.

1378: 

1379: \subsection{Impact on Results}

1380: Now we examine whether our utilization of a single pointing of each dense

1381: \nh core introduces significant bias or error in the core properties

1382: measured.

1383: 

1384: \subsubsection{Line Widths}

1385: 

1386: We calculated the difference in \nh linewidth between each centre position

1387: and the surrounding cross and found this difference to be very small.

1388: Most (93\%) of the \nh cores have mean differences less than one or two spectral

1389: channels (0.05~km/s in FWHM units), i.e., consistent with measuring

1390: the same value.  The mean absolute difference of the entire sample was 

1391: 0.018~km/s and none had differences close to

1392: the sound speed (FWHM of 0.15~km/s for \nhns) and the mean

1393: difference was approximately zero.  Therefore, we expect 

1394: that our analysis utilizing only central pointings does not introduce 

1395: significant error or bias to our measurement of core linewidth.

1396: 

1397: \subsubsection{Centroid Velocities}

1398: The difference in centroid velocity for the \nh cores at centre and

1399: cross positions is also small, as shown in Figure~\ref{fig_core_core_cross}.

1400: In Figure~\ref{fig_core_core_cross}, the vertical lines indicate the

1401: range in differences over the four cross positions.

1402: Note that the six sources with two velocity components

1403: fit at either centre or cross position have not been included in this plot.

1404: All but one of the \nh cores have a difference between the central and average

1405: cross velocity of less than the thermal velocity dispersion; 

1406: the mean difference is approximately zero and the standard deviation

1407: is 0.04~km~s$^{-1}$.  The \nh cores which have

1408: the largest centroid velocity differences have less secure fits where 

1409: there is greater uncertainty in the centroid velocity determined.

1410: On the other hand, in the majority of cases

1411: the difference in velocities between cross and central positions

1412: are larger than the the fitting errors of those velocities,

1413: indicating that there is real variation in the centroid velocity

1414: between locations.  This variation is much smaller

1415: than the core-to-core variations analyzed in the \S6, however,

1416: where extinction regions require support against gravity with

1417: core-to-core motions many times larger than the sound speed.

1418: We thus find that little error or bias is introduced into our results

1419: by using data from a central pointing only.

1420: 

1421: 

1422: %%%%%%%%%%%%%%%%%%%%%%%%%

1423: \begin{thebibliography}{}

1424: \bibitem[Aikawa et al.(2003)]{Aikawa03} Aikawa, Y., Ohashi, N. \& Herbst, E.

1425: 	2003, ApJ, 593, 906

1426: \bibitem[Alves \& Lombardi(2007)]{Alves} Alves, J.\& Lombardi, M.\ 2007,

1427:         in preparation

1428: \bibitem[Ayliffe et al.(2007)]{Ayliffe06} Ayliffe, B.~A., Langdon, J.~C., 

1429:         Cohl, H.~S., \& Bate, M.R. 2007, MNRAS, 374, 1198

1430: \bibitem[Bally et al.(in prep)]{Bally_handbook} Bally, J., 

1431: 	Walawender, J., Johnstone, D., Kirk, H., Goodman, A.,

1432: 	Porras, A. \& Ridge, N. Handbook of Star Forming Regions Vol. I.

1433: 	The Northern Sky (ed. Reipurth), in prep.

1434: \bibitem[Barranco \& Goodman(1998)]{Barranco98} Barranco, J.~A. \&

1435: 	Goodman, A.~A. 1998, ApJ, 504, 207

1436: \bibitem[Bate et al.(2003)]{Bate03} Bate, M.~R., Lubow, S.~H.,

1437: 	Ogilvie, G.~I. \& Miller, K.~A. 2003, MNRAS, 341, 213

1438: \bibitem[Benson \& Myers(1989)]{Benson89} Benson, P.~J. \& Myers, P.~C.

1439: 	1989, ApJS, 71, 89

1440: \bibitem[Bertoldi \& McKee(1992)]{Bertoldi92} Bertoldi, F. \& McKee, C.~F.

1441: 	1992, ApJ, 395, 140

1442: \bibitem[Caselli et al.(1995)]{Caselli95} Caselli, P., Myers, P.~C., 

1443: 	Thaddeus, P. 1995, ApJ, 455L, 77

1444: \bibitem[Caselli \& Myers(1995)]{Caselli95b} Caselli, P. \& Myers, P.~C.

1445: 	1995, ApJ, 446, 665

1446: \bibitem[Caselli et al.(2002)]{Caselli02} Caselli, P., Benson, P.~J.,

1447: 	Myers, P.~C., \&Tafalla, M. 2002, ApJ, 572, 238

1448: \bibitem[${\rm \breve{C}}$ernis(1993)]{Cernis93} ${\rm \breve{C}}$ernis, K. 

1449:         1993, BaltA, 2, 214

1450: \bibitem[Cohen \& Kuhi (1979)]{Cohen79} Cohen, M. \&Kuhi, L.~V. 1979,

1451: 	ApJS, 41, 743

1452: \bibitem[Crapsi et al.(2005)]{Crapsi05} Crapsi, A., Caselli, P., 

1453: 	Walmsley, C.~M., Myers, P.~C., Tafalla, M., Lee, C.~W., 

1454: 	\& Bourke, T.~L. 2005, ApJ, 619, 379

1455: \bibitem[de~Gregorio-Monsalvo et al.(2006)]{deGregorio06} de~Gregorio-Monsalvo,

1456: 	I., G\'omez, J.~F., Su\'arez, O., Kuiper, T.~B.~H., 

1457: 	Rodr\'iguez, L.~F., Jim\'enez-Bail\'on, E. 2006, ApJ, 642, 319

1458: \bibitem[Di~Francesco et al.(2007)]{DiFrancesco07} Di~Francesco, J.,

1459: 	Johnstone, D., Kirk, H., Ledwosinksa, E. \& MacKenzie, T., ApJS submitted

1460: \bibitem[Dib et al.(2007)]{Dib06} Dib, S., Vazquez-Semadeni, E., Kim, J.,

1461: 	Burkert, A. \& Shadmehri, M. 2007, ApJ, 661, 262

1462: \bibitem[Evans et al.(2003)]{Evans03} Evans, N.~J. et al.\ 2003, PASP, 115, 965

1463: \bibitem[Goodman et al.(1998)]{Goodman98} Goodman, A.~A., Barranco, J.~A.,

1464: 	Wilner, D.~J. \& Heyer, M.~H. 1998, ApJ, 504, 223

1465: \bibitem[Hartmann(1998)]{Hartmann98} Hartmann, L.\ 1998, Accretion Processes

1466: 	in Star Formation (Cambridge: Cambridge University Press)

1467: \bibitem[Hatchell et al.(2005)]{Hatchell05} Hatchell, J., Richer, J.~S.,

1468:         Fuller, G.~A., Qualtrough, C.~J., Ladd, E.~F., \& Chandler, C.~J.

1469:         2005, A\&A, 440, 151

1470: \bibitem[Hatchell et al.(2007)]{Hatchell07} Hatchell, J., Fuller, G.~A.,

1471: 	Richer, J.~S., Harries, T.~J., \& Ladd, E.~F. 2007, A\&A, accepted

1472: \bibitem[Holland et al.(1999)]{Holland99} Holland, W.~S., et al.\ 1999, MNRAS,

1473:         303, 659

1474: \bibitem[Jijina et al.(1999)]{Jijina99} Jijina, J., Myers, P.~C., Adams, F.~C.

1475: 	1999, ApJS, 125, 161

1476: \bibitem[Johnstone et al.(2000)]{Johnstone00} Johnstone, D., Wilson, C.D.,

1477:         Moriarty-Schieven, G., Giannakopoulou-Creighton, J., \& Gregersen, E.

1478:         \ 2000, ApJ, 131, 505

1479: \bibitem[Johnstone et al.(2001)]{Johnstone01} Johnstone, D., Fich, M.,

1480:         Mitchell, G.F. \& Moriarty-Schieven, G. 2001, ApJ, 559, 307

1481: \bibitem[Johnstone \& Bally(2006)]{Johnstone06} Johnstone, D. \&

1482: 	Bally, J. 2006, ApJ, 653, 383

1483: \bibitem[J\o rgensen et al.(2006)]{Jorgensen06} J\o rgensen, J.~K., Harvey, P.~M.,

1484: 	Evans, N.~J~II, Huard, T.~L., Allen, L.~E., Porras, A., et al

1485: 	2006, ApJ, 645, 1246

1486: \bibitem[J\o rgensen et al.(2007)]{Jorgensen07} 

1487:         J\o rgensen, J.~K., Johstone, D., Kirk, H., \& Myers, P.~C. 2007,

1488:         ApJ, 656, 293

1489: \bibitem[Kirk, Johnstone, \& Di Francesco(2006)]{Kirk06} Kirk, H., 

1490:         Johnstone, D., \& Di Francesco, J. 2006, ApJ, 646, 1009

1491: \bibitem[Klessen et al.(2005)]{Klessen05} Klessen, R.~S., Ballesteros-Paredes,

1492:         J., V\'azquez-Semadeni, E. \& Dur\'an-Rojas, C. 2005, ApJ, 620, 786

1493: \bibitem[Larson(1981)]{Larson81} Larson, R.~B. 1981, MNRAS, 194, 809

1494: \bibitem[Lee, Myers \& Tafalla(2001)]{Lee01} Lee, C.~W., Myers, P.~C.,

1495: 	\& Tafalla, M. 2001, ApJS, 136, 703

1496: \bibitem[Lombardi \& Alves(2001)]{Lombardi01} Lombardi, M. \& Alves, J.

1497:         2001, A\&A, 377, 1023

1498: \bibitem[Morata et al.(2005)]{Morata05} Morata, O., Girart, J.~M.,

1499: 	Estalella, R. 2005, A\&A, 435, 113

1500: \bibitem[M\"uller et al.(2001)]{Muller01} M\"uller, H.~S.~P., Thorwirth, S., 

1501: 	Roth, D.~A. \& Winnewisser, G. 2001, A\&A, 370, L49 

1502: \bibitem[Rebull et al.(in press)]{Rebull07} Rebull, L. et al., 2007, 

1503: 	ApJ in press

1504: \bibitem[Ridge et al.(2006)]{Ridge06} Ridge, N.~A., Di~Francesco, J., Kirk, H.,

1505: 	Li, D., Goodman, A.~A., et al. 2006, AJ, 131, 2921

1506: \bibitem[Shirley et al.(2005)]{Shirley05} Shirley, Y.~L., Nordhaus, M.~K.,

1507: 	Grcevich, J.~M., Evans, N.~J.~II, Rawlings, J.~M.~C. \&

1508: 	Tatematsu, K. 2005, ApJ, 632, 982

1509: \bibitem[Tafalla et al.(2002)]{Tafalla02} Tafalla, M., Myers, P.~C., 

1510: 	Caselli, P., Walmsley, C.~M., Comito, C. 2002, ApJ, 569, 815

1511: \bibitem[Tafalla et al.(2004)]{Tafalla04} Tafalla, M., Myers, P.~C.,

1512: 	Caselli, P., \& Walmsley, C.~M. 2004, A\&A, 416, 191

1513: \bibitem[Tafalla et al.(2006)]{Tafalla06} Tafalla, M., Kumar, M.~S.~N.,

1514: 	\& Bachiller, R. 2006, A\&A, 456, 179

1515: \bibitem[Ungerechts et al.(1997)]{Ungerechts97}  Ungerechts, H., 

1516: 	Bergin, E.~A., Goldsmith, P.~F., Irvine, W.~M., Schloerb, F.~P.

1517: 	\& Snell, R.~L. 1997, ApJ, 482, 245

1518: \bibitem[Walawender et al.(2005)]{Walawend05} Walawender, J., Bally, J.,

1519:         Kirk, H., Johnstone, D.\ 2005, AJ, 130, 1795

1520: \bibitem[Walawender et al.(2006)]{Walawend06} Walawender, J., Bally, J.,

1521:         Kirk, H., Johnstone, D., Reipurth, B., \& Aspin, C.\ 2006,

1522: 	AJ, 132, 467

1523: \bibitem[Walsh et al.(2004)]{Walsh04} Walsh, A.~J., Myers, P.~C. \&

1524: 	Burton, M.~G. 2004, ApJ, 614, 194

1525: \bibitem[Walsh et al.(2007)]{Walsh06} Walsh, A.~J., Myers, P.~C., 

1526: 	Di Francesco, J., Mohanty, S., Bourke, T.~L., Gutermuth, R.,

1527: 	\& Wilner, D. 2007, ApJ, 655, 958

1528: \bibitem[Ward-Thompson et al.(2006)]{Wardthomp06}Ward-Thompson, D., 

1529: 	Nutter, D, Bontemps, S, Whitworth, A, \& Attwood, R. 2006, MNRAS,

1530: 	369, 1201

1531: \bibitem[Ward-Thompson et al.(2007)]{Wardthomp07} Ward-Thompson, D.,

1532: 	Andr\'e, P., Crutcher, R., Johnstone, D., Onishi, T. \&

1533: 	Wilson, C. 2007, in Protostars and Planets V, ed. B. Reipurth, 

1534: 	D. Jewitt, \& K. Keil (Tucson: Univ. Arizona Press), 33

1535: \bibitem[Williams, de Geus, \& Blitz(1994)]{Williams94} Williams, J.P.,

1536:         de Geus, E.J., \& Blitz, L.\ 1994, ApJ, 428, 693

1537: \end{thebibliography}

1538: 

1539: %%%%%%%%%%%%%%%%%%%%%%%%

1540: \input{tab1}

1541: \input{tab2}

1542: \input{tab3}

1543: \input{tab4}

1544: \input{tab5}

1545: \input{tab6}

1546: 

1547: %%%%%%%%%%%%%%%%%%%%%%%

1548: %figure of targets / detections

1549: \clearpage

1550: \begin{figure}[p]

1551: \includegraphics[height=11.5cm,angle=90]{f1_color.ps}

1552: %\plotone{f1_color.ps}

1553:   \caption{2MASS-derived extinction map of the Perseus molecular cloud 

1554: 	overlaid with the positions of our IRAM survey targets.  Red

1555: 	circles indicate the SCUBA-selected targets, green triangles

1556: 	indicate the Palomar plate-selected targets, and blue squares

1557: 	indicate the 2MASS-selected targets.  Well-known star-formation

1558: 	regions are labelled.

1559:         }

1560: %Caption for b/w paper version

1561: %  \caption{{2MASS-derived extinction map of the Perseus molecular cloud

1562: %	overlaid with the positions of our IRAM survey targets.  Medium grey

1563: %	circles indicate the SCUBA-selected targets, light grey triangles

1564: %	indicate the Palomar plate-selected targets, and dark grey squares

1565: %	indicate the 2MASS-selected targets.  Well-known star-formation

1566: %	regions are labelled.

1567: %	}

1568: 

1569:   \label{fig_targets}

1570: \end{figure}

1571: 

1572: %%Sample CO Spectra

1573: \clearpage

1574: \begin{figure}

1575: %\includegraphics[height=11.5cm]{f2_color.ps}

1576: \plotone{f2_color.ps}

1577:   \caption{Three example \co spectra showing differing spectral

1578: 	profiles.  Black indicates the data while the red

1579: 	indicates the model fit.  Blue and green indicate the

1580: 	components of a two Gaussian model.  The vertical axis

1581: 	is in units of T$_{A}^{*}$.}

1582: %Caption for the paper b/w version

1583: %  \caption{Three example \co spectra showing differing spectral

1584: %        profiles.  Black indicates the data while dark grey

1585: %        indicates the model fit.  The lighter grey lines indicate the

1586: %        components of a two Gaussian model.  The vertical axis

1587: %        is in units of T$_{A}^{*}$.}

1588: 

1589:   \label{fig_sample_co}

1590: \end{figure}

1591: 

1592: %the internal turb figure for nh

1593: \begin{figure}

1594: %\includegraphics[height=11cm]{f3_color.ps}

1595: \plotone{f3_color.ps}

1596:   \caption{Relative level of non-thermal motions within the dense

1597: 	\nh cores.  The top horizontal axis shows the observed FWHM 

1598: 	linewidth in km/s while the bottom axis shows the turbulent fraction 

1599: 	f$_{turb}$ assuming a temperature of 15~K.

1600: 	The top panel shows the protostars (solid green

1601: 	line) and starless cores (dashed green line).  The bottom panel

1602: 	shows the starless cores (dashed green line) versus the

1603: 	prediction from a gravoturbulent simulation by \citet{Klessen05} for

1604: 	starless cores (dotted red line).

1605: 	Note that the final histogram bin for the \citet{Klessen05} model 

1606: 	includes all objects above this turbulent fraction (which extends

1607: 	to 4.3 in their model). 

1608:         }

1609: %Caption for paper b/w version

1610: %  \caption{Relative level of non-thermal motions within the dense

1611: %        \nh cores.  The top horizontal axis shows the observed FWHM

1612: %        linewidth in km/s while the bottom axis shows the turbulent fraction

1613: %        f$_{turb}$ assuming a temperature of 15~K.

1614: %        The top panel shows the protostars (solid light grey

1615: %        line) and starless cores (dashed dark grey line).  The bottom panel

1616: %        shows the starless cores (dashed dark grey line) versus the

1617: %        prediction from a gravoturbulent simulation by \citet{Klessen05} for

1618: %        starless cores (dotted light grey line).

1619: %        Note that the final histogram bin for the \citet{Klessen05} model

1620: %        includes all objects above this turbulent fraction (which extends

1621: %        to 4.3 in their model).

1622: %        }

1623: 

1624:   \label{fig_internal_turb_nh}

1625: \end{figure}

1626: 

1627: %figure of internal turbulence level for CO 

1628: \begin{figure}

1629: %\includegraphics[height=11cm]{f4_color.ps}

1630: \plotone{f4_color.ps}

1631:   \caption{Relative level of non-thermal motions measured in \cons.

1632: 	The top horizontal axis shows the observed FWHM linewidth

1633: 	in km/s while the bottom axis shows the turbulent fraction 

1634: 	f$_{turb}$ assuming a temperature of 15~K.

1635: 	The top panel shows the targets associated with protostars

1636: 	(solid blue line) and those not associated with protostars (dashed

1637: 	blue line).  The bottom panel shows the targets not associated

1638: 	with protostars (dashed blue line)

1639: 	versus the prediction from a gravoturbulent simulation by 

1640: 	\citet{Klessen05} (dotted red line).

1641: 	Note that the final histogram bin for the \citet{Klessen05} model 

1642: 	includes all objects above this turbulent fraction (which extends

1643: 	to 4.3 in their model). 

1644: 	}

1645: %Caption for b/w paper version

1646: %  \caption{Relative level of non-thermal motions measured in \cons.

1647: %        The top horizontal axis shows the observed FWHM linewidth

1648: %        in km/s while the bottom axis shows the turbulent fraction

1649: %        f$_{turb}$ assuming a temperature of 15~K.

1650: %        The top panel shows the targets associated with protostars

1651: %        (solid light grey line) and those not associated with protostars 

1652: %	 (dashed dark grey

1653: %        line).  The bottom panel shows the targets not associated

1654: %        with protostars (dashed dark grey line)

1655: %        versus the prediction from a gravoturbulent simulation by

1656: %        \citet{Klessen05} (dotted light grey line).

1657: %        Note that the final histogram bin for the \citet{Klessen05} model

1658: %        includes all objects above this turbulent fraction (which extends

1659: %        to 4.3 in their model).

1660: %        }

1661: 

1662:   \label{fig_internal_turb_co}

1663: \end{figure}

1664: 

1665: %%Core vs Envelope (single CO vel)

1666: \begin{figure}

1667: %\includegraphics[height=11cm]{f5.ps}

1668: \plotone{f5.ps}

1669:   \caption{Difference in centroid veloicites of \nh to \co 

1670: 	for the starless cores (left) and protostars (right).  The

1671: 	dotted lines indicate the sound speed of the ambient medium.

1672: 	The dashed lines indicate Gaussian fits to the distributions 

1673: 	-- the starless cores have $\sigma =$ 0.17~km/s while the

1674: 	protostars have $\sigma =$ 0.16~km/s.}

1675:   \label{fig_core_rel_envelope_1}

1676: \end{figure}

1677: 

1678: %%Core vs Envelope (double CO vel)

1679: \begin{figure}

1680: %\includegraphics[height=11cm]{f6.ps}

1681: \plotone{f6.ps}

1682:   \caption{Difference in centroid velocities of \nh and \co for

1683: 	starless cores (top) and protostars (bottom).  The closest \co

1684: 	velocity components are denoted by diamonds or asterisks while 

1685: 	the farther

1686: 	\co velocity components are denoted by squares.  In each instance,

1687: 	the cores are ordered from largest to smallest difference in 

1688: 	velocity using the farther \co velocity component.  The dotted

1689: 	line indicates the sound speed in the ambient medium.}

1690:   \label{fig_core_rel_envelope_2}

1691: \end{figure}

1692: 

1693: %figure - core-to-core velocity dispersion

1694: \begin{figure}

1695: %\includegraphics[height=11cm]{f7_color.ps}

1696: \plotone{f7_color.ps}

1697:   \caption{Ratio of the measured velocity dispersion to that required to

1698: 	counteract gravity ($\sqrt{GM_{ext}/5R_{ext}}$)

1699: 	versus the mass in the extinction region.

1700: 	The dotted line shows the expected relationship for virial equilibrium.

1701: 	The green filled diamonds denote the dispersion in centroid velocity for

1702: 	\nh starless cores.  The blue open diamonds denote the dispersion of

1703: 	the summed \co spectra for all starless cores in the extinction 

1704: 	region.  Two extinction regions have no \nh dispersion

1705: 	measured since less than two \nh cores were detected. }

1706: 	%Note for me: From left to right, the extinction regions correspond to 

1707: 	%\#11, 10, 9, 8, 6, 5, 4, 3, 7, 1, and 2. }

1708: %Caption for b/w paper version

1709: %  \caption{Ratio of the measured velocity dispersion to that required to

1710: %        counteract gravity ($\sqrt{GM_{ext}/5R_{ext}}$)

1711: %        versus the mass in the extinction region.

1712: %        The dotted line shows the expected relationship for virial equilibrium.

1713: %        The filled diamonds denote the dispersion in centroid velocity for

1714: %        \nh starless cores.  The open diamonds denote the dispersion of

1715: %        the summed \co spectra for all starless cores in the extinction

1716: %        region.  Two extinction regions have no \nh dispersion

1717: %        measured since less than two \nh cores were detected. }

1718: 

1719:   \label{fig_core_core}

1720: \end{figure}

1721: 

1722: %Figure - concentration vs velocity dispersion

1723: \begin{figure}

1724: %\includegraphics[height=11cm]{f8.ps}

1725: \plotone{f8.ps}

1726:   \caption{Variation of observed velocity dispersion in \nh cores

1727: 	versus concentration for starless cores (diamonds) and protostellar 

1728: 	cores (asterisks).  The large bold crosses indicate the

1729: 	mean and standard deviation for each of the two samples.

1730: 	The smaller symbols indicate SCUBA sources for which the properties

1731: 	derived from clumpfind are less secure (see discussion in 

1732: 	Appendix A).  These sources were not used in any of the 

1733: 	calcuations.}

1734:   \label{fig_internal_turb_concs}

1735: \end{figure}

1736: 	

1737: %figure - int turb vs mass

1738: \begin{figure}

1739: %\includegraphics[height=11cm]{f9.ps}

1740: \plotone{f9.ps}

1741:   \caption{Variation of core velocity dispersion with the

1742: 	total flux within a SCUBA core for both protostars and

1743: 	starless cores.  The same plotting convention is used

1744: 	as in Figure~\ref{fig_internal_turb_concs}.

1745: 	}

1746:   \label{fig_internal_turb_flux}

1747: \end{figure}

1748: 

1749: %figure - flux vs ratio of virial to observed mass

1750: \clearpage

1751: \begin{figure}

1752: %\includegraphics[height=11cm]{f10.ps}

1753: \plotone{f10.ps}

1754:   \caption{Ratio of the mean gas velocity dispersion to the

1755: 	virial velocity dispersion (calculated from SCUBA core flux and

1756: 	radius) squared, versus the observed SCUBA flux.  The same plotting

1757: 	conventions are used as in Figure~\ref{fig_internal_turb_concs}.

1758: 	The dotted line shows the expected relationship for virial equilibrium.

1759: 	The dashed and dot-dashed lines show the relationship for thermal

1760: 	15~K cores assuming core radii of 10\arcsec\ and 60\arcsec\ 

1761: 	respectively.} 

1762:   \label{fig_internal_turb_mass}

1763: \end{figure}

1764: %%figure - integrated intensity vs flux

1765: \begin{figure}

1766: %\includegraphics[height=11cm]{f11.ps}

1767: \plotone{f11.ps}

1768:   \caption{Variation in integrated intensity for \co and \nh

1769: 	with total SCUBA flux over the region observed by IRAM.  

1770: 	Note that cores associated with total

1771: 	SCUBA fluxes of over 8~Jy were included in the plot

1772: 	as having SCUBA fluxes of 8~Jy~beam$^{-1}$.  The same plotting

1773: 	conventions are used as in Figure~\ref{fig_internal_turb_concs}.

1774: 	The squares with vertical lines indicate the mean size of the

1775: 	error in the integrated intensity.  The error bar for the

1776: 	\co integrated intensity has been enlarged by a factor of two.

1777: 	}

1778:   \label{fig_intens_vs_submm}

1779: \end{figure}

1780: 

1781: %figure - freezeout? (ratio nh & CO intensities)

1782: \begin{figure}

1783: %\includegraphics[height=11cm]{f12.ps}

1784: \plotone{f12.ps}

1785:   \caption{Ratio in \co to \nh integrated intensity versus total SCUBA

1786: 	flux over the IRAM beam.  Note that all cores associated with 

1787: 	SCUBA fluxes of over 8~Jy have been included in the plot as 

1788: 	having values of 8~Jy.  The diamonds denote starless cores while 

1789: 	the asterisks denote the protostars.}

1790:   \label{fig_intens_ratio_vs_submm}

1791: \end{figure}

1792: %figure - column density dust vs nh comparison

1793: \begin{figure}

1794: %\includegraphics[height=11cm]{f13.ps}

1795: \plotone{f13.ps}

1796:   \caption{Column density of \nh versus the total column density 

1797: 	(calculated using the SCUBA flux within the IRAM beam).

1798: 	The diagonal lines indicate fractional abundances of \nhns. 

1799: 	Starless cores are diamonds while protostars are asterisks.}

1800: \label{fig_coldens}

1801: \end{figure}

1802: 

1803: %% Appendix figures - doublefit regions.

1804: \begin{figure}

1805: %\includegraphics[height=18cm]{f14.ps}

1806: \plotone{f14.ps}

1807:   \caption{The 3\arcsec\ SCUBA map of the B1 star forming region.

1808: 	The image is scaled such that white corresponds to 

1809: 	$\sim$0~Jy~beam$^{-1}$and black to $\sim$0.25~Jy~beam$^{-1}$.}

1810: %now B1

1811: \label{fig_submm_ngc1333}

1812: \end{figure}

1813: 

1814: \begin{figure}

1815: %\includegraphics[height=11cm]{f15_color.ps}

1816: \plotone{f15_color.ps}

1817:    \caption{Spectrum for source \#99 in NGC1333.  The vertical axis is in

1818: 	units of T$_{A}^{*}$.  Red indicates the summed fit while blue

1819: 	and green indicate the two different components fit (where applicable).}

1820: %Caption for b/w paper version

1821: %   \caption{Spectrum for source \#99 in NGC1333.  The vertical axis is in

1822: %        units of T$_{A}^{*}$.  Dark grey indicates the summed fit while 

1823: %	 lighter shades indicate the two different components fit (where 

1824: %	 applicable).}

1825:    \label{fig_ps_s36}

1826: \end{figure}

1827: 

1828: \begin{figure}

1829: %\includegraphics[height=11cm]{f16_color.ps}

1830: \plotone{f16_color.ps}

1831:   \caption{Spectrum for source \#27 in IC348.  The same plotting convention

1832: 	is used as in Figure~\ref{fig_ps_s36}.}

1833: \label{IC348}

1834: \end{figure}

1835: 

1836: \begin{figure}

1837: %\includegraphics[height=11cm]{f17_color.ps}

1838: \plotone{f17_color.ps}

1839:   \caption{Spectrum for source \#136 in L1544.  The same plotting convention

1840: 	is used as in Figure~\ref{fig_ps_s36}.}

1841: \label{fig_ps_s41}

1842: \end{figure}

1843: 

1844: \begin{figure}

1845: %\includegraphics[height=11cm]{f18_color.ps}

1846: \plotone{f18_color.ps}

1847:   \caption{Spectrum of source \#148, the two velocity component core in L1448.

1848: 	The same plotting convention is used as in Figure~\ref{fig_ps_s36}.}

1849: \label{L1448}

1850: \end{figure}

1851: 

1852: %figure linewidth variation across offsets.

1853: \begin{figure}

1854: %\includegraphics[height=11cm]{f19.ps}

1855: \plotone{f19.ps}

1856:    \caption{Fractional difference in integrated intensity between 

1857: 	centre and offset positions in dense \nh cores.  The pluses denote the

1858: 	maximum difference with the vertical lines indicating the error

1859: 	in this measurement.  The squares denote the average difference

1860: 	for all four offset positions versus the centre.

1861: 	The cores are ordered according to the maximum difference.}

1862:    \label{fig_cross_area}

1863: \end{figure}

1864: 

1865: %figure of cross vs centre velocities

1866: \begin{figure}

1867: %\includegraphics[height=11cm]{f20.ps}

1868: \plotone{f20.ps}

1869:   \caption{Difference in centroid velocity between the cross and central

1870: 	positions, with \nh cores ordered in increasing average difference.  

1871: 	The diamonds indicate the average cross velocity difference

1872: 	while the vertical lines indicate the range in centroid velocity 

1873: 	difference at all cross positions.  Cores with two velocity

1874: 	components fit have not been included.  The dotted line

1875: 	indicates the sound speed.

1876: 	}

1877:   \label{fig_core_core_cross}

1878: \end{figure}

1879: 

1880: 

1881: \end{document}

1882: