astro-ph0002446/ms.tex
1: %MS#50721
2: %\input espf
3: \documentstyle[12pt,aasms4]{article}
4: %\documentstyle[10pt,emulateapj]{article}
5: 
6: \received{November 3 1999}
7: \accepted{February 23 2000}
8: %\journalid{337}{00 October 1996}
9: %\articleid{11}{14}
10: 
11: %
12: 
13: \begin{document}
14: 
15: \title{The $L_x$-$T$, $L_x$-$\sigma$ and 
16: $\sigma$-$T$ relations for groups and clusters of galaxies}
17: 
18: \author{Yan-Jie Xue and Xiang-Ping Wu}
19: 
20: \affil{Beijing Astronomical Observatory and 
21:        National Astronomical Observatories, 
22:        Chinese Academy of Sciences, Beijing 100012, China}
23: 
24: \begin{abstract}
25: While in the hierarchical model of structure formation,
26: groups of galaxies are believed to be the scaled-down version of clusters
27: of galaxies, a similarity breaking in the fundamental laws 
28: may occur on the  group scale, 
29: reflecting a transition between galaxy-dominated and intracluster
30: medium dominated properties. In this paper, 
31: we present an extensive study of 
32: the relations between the X-ray luminosity ($L_x$), 
33: the temperature ($T$) of hot diffuse gas and the velocity 
34: dispersion ($\sigma$) of galaxies for groups and clusters of galaxies, 
35: based on the largest sample of 66 groups and 274 clusters drawn from 
36: literature. Our best fit $L_x$-$T$ and $L_x$-$\sigma$ relations for groups
37: read $L_x\propto T^{5.57\pm1.79}\propto \sigma^{2.35\pm0.21}$, which
38: deviates remarkably from those for clusters: 
39: $L_x\propto T^{2.79\pm0.08}\propto \sigma^{5.30\pm0.21}$.
40: The significance of these correlations have been justified
41: by both the co-consistency test and the Kendall's $\tau$ statistics.   
42: We have thus confirmed the existence of 
43: similarity breaking in the $L_x$-$T$ and $L_x$-$\sigma$ relations
44: between groups and clusters as claimed in previous work,
45: although the best fit $\sigma$-$T$ relations remain roughly the same
46: in both systems: $\sigma\propto T^{0.64}$. Alternatively,  
47: the significant disagreement between the observationally
48: fitted $L_x$-$T$ and $L_x$-$\sigma$ relations for groups and 
49: those expected from a perfect hydrostatic equilibrium hypothesis 
50: indicates that the X-ray emission of individual galaxies and 
51: the non-gravitational heating 
52: must play a potentially important role in the dynamical  
53: evolution of groups. Therefore, reasonable caution should be exercised in
54: the cosmological applications of the dynamical properties of groups.
55: \end{abstract}
56: 
57: \keywords{cosmology: observations --- galaxies: clusters: general ---  
58:           X-rays: galaxies}
59: 
60: 
61: \section{Introduction}
62: 
63: 
64: It has been well established that there exists a strong correlation
65: between the X-ray determined bolometric luminosity $L_x$,  
66: the X-ray temperature $T$ of the intracluster gas 
67: and the optical measured velocity dispersion
68: $\sigma$ of the cluster galaxies (Wu, Xue \& Fang 1999; hereafter Paper I
69: and references therein). 
70: A precise determination of these correlations is important 
71: not only for the study of dynamical properties and evolution of clusters
72: themselves but also for distinguishing different cosmological models 
73: including the prevailing estimate of the cosmic density parameter $\Omega_M$ 
74: through a combination of the baryon fraction  of clusters and 
75: the Big Bang Nucleosynthesis (White et al. 1993;
76: David, Jones \& Forman 1995). 
77: For instance, if the observed $L_x$-$T$ relation follows
78: $L_x\propto T^{2}$ (e.g. Quintana \& Melnick 1982; Markevitch et al. 1998), 
79: we would arrive at the conclusion that the X-ray 
80: emission is primarily due to purely gravitational heating and 
81: thermal bremsstrahlung, and the baryon fraction $f_b$  of clusters can 
82: be representative of the Universe,
83: i.e., $f_b$ provides  a robust estimate of $\Omega_M$ (see Paper I). 
84: However, if the observed $L_x$-$T$ relation appears to be $L_x\propto T^{3}$
85: (e.g. White, Jones \& Forman 1997), other non-gravitational heating 
86: mechanisms must be invoked in order to give gas sufficient 
87: excess energy (e.g. Ponman, Cannon \& Navarro 1999; 
88: Wu, Fabian \& Nulsen 1999; Loewenstein 1999), unless  
89: the baryon fraction of clusters is requited to vary with
90: temperature (David et al. 1993). The latter challenges the standard model of
91: structure formation.  
92: 
93: 
94: Another critical issue is whether the $L_x$-$T$, $L_x$-$\sigma$ and 
95: $\sigma$-$T$ relations for clusters of galaxies can naturally extend 
96: to group scales. In the hierarchical model of structure formation,
97: groups are believed to be the scaled-down 
98: version of clusters, and the underlying gravitational potentials of groups and
99: clusters are similar when scaled to their virial radii (e.g. Navarro,
100: Frenk \& White 1995). It is expected that groups and clusters should 
101: exhibit similar correlations between $L_x$, $T$ and $\sigma$, 
102: provided that gas and galaxies 
103: are in hydrostatic equilibrium with the underlying gravitational potential
104: of groups and their X-ray emissions are produced by thermal bremsstrahlung.
105: Indeed, in the present Universe  a majority of the baryons may be bound in the 
106: gravitational wells of groups  (Fukugita, Hogan \& Peebles 1998). 
107: All groups may contain hot X-ray emitting gas with 
108: $kT$ around or less than $1$ keV (e.g. Price et al. 1991; Ponman et al. 1996),
109: and most of them should be detectable with future sensitive observations. 
110: Without the knowledge about the dynamical properties of groups 
111: characterized by the $L_x$-$T$, $L_x$-$\sigma$ and $\sigma$-$T$ relations, 
112: it could be misleading if the presently estimated baryon fraction of groups 
113: is immediately used as a cosmological indicator.
114: Meanwhile, any difference in these correlations between 
115: clusters and groups will be helpful for
116: our understanding of the formation and evolution of structures on scales of 
117: $1$--$10$ Mpc and of the significance of non-gravitational 
118: heating mechanisms.  
119: 
120: 
121: The pioneering work of constructing  the  $L_x$-$\sigma$ relation for groups
122: was carried out by Dell'Antonio, Geller \& Fabricant (1994). 
123: They found a significant flattening in the relation for groups with
124: $\sigma$ below $300$ km s$^{-1}$, as compared with the same
125: relation for rich clusters. 
126: An extensive study of the $L_x$-$T$, $L_x$-$\sigma$ 
127: and $\sigma$-$T$ relations for groups was soon made by Ponman et al.(1996), 
128: based on 22 Hickson's compact groups (HCG) whose X-ray emissions are detected.
129: The most remarkable result is the steepening ($L_x\propto T^{8.2}$)
130: of the $L_x$-$T$ relation, in contrast with the X-ray properties of
131: clusters ($L_x\propto T^{2.5}$--$T^{3}$), while the significant flattening
132: in the  $L_x$-$\sigma$ relations for groups claimed by 
133: Dell'Antonio et al. (1994)
134: was only marginally detected. Using the RASSCALS group catalog, 
135: Mahdavi et al. (1997, 2000) obtained a much shallower $L_x$-$\sigma$ 
136: relation ($L_x\propto \sigma^{0.37\pm0.3}$)
137: for groups with $\sigma<340$ km s$^{-1}$ than the above results. 
138: Interestingly, Mulchaey \& Zabludoff (1998)
139: studied 12 poor ROSAT groups but arrived at an opposite conclusion that  
140: the X-ray component in groups follows the same $L_x$-$T$-$\sigma$ relations
141: as those in clusters.
142: Namely, groups are indeed low-mass versions of clusters.
143: 
144: 
145: Theoretically, several investigations have been made on the possible reasons
146: why the $L_x$-$T$ relation flattens with the increase of 
147: scale and temperature (e.g. Cavaliere, Menzi \& Tozzi 1997, 1999; 
148: Valageas \& Schaeffer 1999). 
149: In particular, if the $L_x$-$T$ and $L_x$-$\sigma$ relations show
150: a significant departure from the expectations of 
151: $L_x\propto T^{2}\propto \sigma^{4}$   
152: within the framework of thermal bremsstrahlung and 
153: hydrostatic equilibrium, the previous estimate of the amount of baryonic 
154: matter in groups and its application to the determination of 
155: cosmic density parameter would become questionable.
156: Consequently, one may need to study
157: whether the observed X-ray emission of groups has partially arisen from
158: the individual galaxies ( Dell'Antonio et al. 1994) or
159: other non-gravitational mechanisms such as those suggested by 
160: the preheated ICM model 
161: (e.g. Ponman et al. 1999) and the shock model (Cavaliere et al. 1997). 
162: 
163: 
164: In this paper, we would like to update and then compare  
165: the $L_x$-$T$, $L_x$-$\sigma$ and $\sigma$-$T$ relations for groups
166: and clusters, using all the available X-ray/optical 
167: data in literature especially the new measurements of 
168: poor clusters and groups. We wish to achieve a better statistical 
169: significance, which may essentially allow us to closely examine 
170: the question as to whether the similarity of these relations 
171: would break down on group scales. 
172: Throughout the paper we assume a Hubble constant of
173: $H_0=50$ km s$^{-1}$ Mpc$^{-1}$ and a flat cosmological model of
174: $\Omega_0=1$.
175: 
176: 
177: \section{Sample}
178: 
179: 
180: We follow the same strategy as in Paper I to select
181: groups and clusters of galaxies from literature: We include all the
182: groups and clusters for which 
183: at least two parameters among the X-ray bolometric luminosity ($L_x$) and
184: temperature ($T$), and the optical velocity dispersion of galaxies ($\sigma$)
185: are observationally determined. Essentially, we use the cluster catalog
186: in Paper I, which contains a total of 256 clusters. We first remove
187: MKW9 from the list and put it into our group catalog. We then add another
188: 19 clusters whose temperatures or velocity dispersions have 
189: become available since then. 
190: This is mainly due to the recent temperature measurements
191: by White (1999). Our final cluster sample consists of 274 clusters. 
192: Unlike the optimistic situation for X-ray clusters, the X-ray emission 
193: has remained  undetectable for most groups
194: because of their  relatively low X-ray temperatures. 
195: By extensively searching the literature, we find 66 groups that meet our
196: criteria, which include 23 HCGs (Table 1).  
197: As compared with the sample used in previous similar study 
198: by Ponman et al. (1996),  the number of groups has almost tripled.  
199: We convert the observed X-ray
200: luminosities to bolometric luminosities in the rest frame of the groups
201: and clusters according to an optically thin and isothermal plasma emission
202: model with 30 percent solar abundance, 
203: in which we assume $T=6$ and $1$ keV, respectively, for the 99 clusters
204: and 26 groups whose temperatures remain unknown spectroscopically.
205: 
206: 
207: 
208: %\placetable{tbl-1}
209: \include{table1-included}
210: 
211: \section{Results}
212: 
213: Essentially, two linear regression methods are used in the fit of 
214: the observed data ($L$, $T$ and $\sigma$) to a power-law relation: the standard
215: ordinary least-square (OLS) method and the orthogonal distance regression
216: (ODR) technique (Feigelson \& Babu 1992). The latter is preferred because
217: it can account for data scatters around two variables, which is 
218: particularly suited when two variables contain significant measurement
219: uncertainties. We perform both the OLS and ODR fits to the data sets of 
220: our group and cluster samples, respectively. 
221: In order to maximally use the published data especially 
222: for groups in the ODR fitting, we assign the average values of
223: measurement uncertainties in $L_x$, $T$ and $\sigma$ to those 
224: data whose error bars are not available. Specifically, 
225: the average relative errors ($\Delta L_x/L_x$, $\Delta T/T$,
226: $\Delta\sigma/\sigma$) are found to be ($14.7\%$, $22.4\%$, $16.5\%$)  
227: and ($24.7\%$, $15.5\%$, $28.4\%$) for clusters and groups, respectively.
228: Finally, the Monte-Carlo simulations are performed to estimate the 
229: standard deviations in the fitted relations. 
230: 
231: 
232: 
233: \subsection{$L_x$-$T$ relation}
234: 
235: 
236: The observed and our best-fit $L_x$-$T$ relations are shown in Fig.1
237: and also summarized in Table 2. It appears that the resultant 
238: $L_x$-$T$ relation for (184) clusters remains nearly the same as that given
239: in Paper I for 142 clusters, $L_x\propto T^{2.79\pm0.08}$.
240: However, our best fit  
241: $L_x$-$T$ relation for 38 groups becomes somewhat flatter: The power-index 
242: in the ODR fitting drops to $5.57\pm1.79$, in contrast to the value of
243: $8.2\pm2.7$ reported by Ponman et al. (1996) based on 16 HCGs.
244: Nevertheless, at $3\sigma$ significance level we have confirmed 
245: their claim for the similarity breaking of the $L_x$-$T$ relation
246: at the low temperature end.
247: 
248:  \begin{deluxetable}{ccllll}
249:  \tablewidth{35pc}
250:  \scriptsize
251:  \tablecaption{The best fit $L_x$-$T$ relations$^a$}
252:  \tablehead{
253:  \colhead{sample}& \colhead{number} &
254:  \colhead{OLS} & 
255:  \colhead{ODR} &
256:  \colhead{$\tau$} &
257:  \colhead{$P$} }
258:  \startdata
259:  group   & 38  &    $L_x=10^{-0.39\pm0.19}T^{2.10\pm0.10}$ &
260:                     $L_x=10^{-0.27\pm0.05}T^{5.57\pm1.79}$ 
261:                &    $0.460$ &  $4.852\times10^{-5}$ \nl
262: cluster  & 184 &    $L_x=10^{-0.89\pm0.08}T^{2.54\pm0.11}$ &
263:                     $L_x=10^{-0.032\pm0.065}T^{2.79\pm0.08}$ 
264:                &    $0.686$ &  $1.569\times10^{-43}$ \nl
265: combined & 222 &    $L_x=10^{-0.17\pm0.19}T^{2.85\pm0.04}$ &
266:                     $L_x=10^{-0.14\pm0.05}T^{3.03\pm0.06}$ 
267:                &    $0.763$ &  $0$ \nl
268: \tablenotetext{a}{$L_x$ and $T$ are in units of $10^{42}$ erg s$^{-1}$ and keV,
269: 	   respectively.}
270:  \enddata
271:  \end{deluxetable}
272: 
273: 
274: \placefigure{fig1}
275: 
276: \subsection{$L_x$-$\sigma$ relation}
277: 
278: 
279: We display in Fig.2 and Table 3 the X-ray luminosity-velocity dispersion
280: relations for 59 groups and 197 clusters. Again, the $L_x$-$\sigma$
281: relation for clusters agrees with our previous result based on 156 clusters
282: (Paper I): $L_x\propto\sigma^{5.3}$, while the best fit 
283: $L_x$-$\sigma$ relation for our group sample  
284: reads $L_x\propto \sigma^{2.35\pm0.21}$, which
285: is significantly flatter than both the above relation for clusters and 
286: the Ponman et al. (1996) result for 22 HCGs 
287: $L_x\propto\sigma^{4.9\pm2.1}$. 
288: Yet, our $L_x$-$\sigma$ relation for groups has not reached 
289: the shallow slopes (0.37 -- 1.56) reported by Mahdavi et al. (1997, 2000).
290: A glimpse of Fig.2 reveals the following two features: (1)The data of 
291: groups and clusters are joined together through poor cluster 
292: population, and there is no apparent gap in between; 
293: (2)The scatters of $\sigma$ around the best fit $L_x$-$\sigma$ relation
294: become relatively large on group scale. 
295: 
296: 
297: 
298:  \begin{deluxetable}{ccllll}
299:  \tablewidth{35pc}
300:  \scriptsize
301:  \tablecaption{The best fit $L_x$-$\sigma$ relations$^a$}
302:  \tablehead{
303:  \colhead{sample}& \colhead{number} &
304:  \colhead{OLS} & 
305:  \colhead{ODR} &
306:  \colhead{$\tau$} &
307:  \colhead{$P$} }
308:  \startdata
309:  group   & 59  &    $L_x=10^{-2.95\pm0.30}\sigma^{1.00\pm0.12}$ &
310:                     $L_x=10^{-6.38\pm0.55}\sigma^{2.35\pm0.21}$ 
311:                &    $0.280$ &  $1.720\times10^{-3}$ \nl
312: cluster  & 197 &    $L_x=10^{-7.18\pm0.55}\sigma^{2.72\pm0.19}$ &
313:                     $L_x=10^{-13.68\pm0.61}\sigma^{5.30\pm0.21}$
314:                &    $0.527$ &  $3.783\times10^{-28}$ \nl 
315: combined & 256 &    $L_x=10^{-8.75\pm0.25}\sigma^{3.55\pm0.06}$ &
316:                     $L_x=10^{-12.10\pm0.52}\sigma^{4.75\pm0.18}$
317:                &    $0.634$ &  $0$ \nl 
318: \tablenotetext{a}{$L_x$ and $\sigma$ are in units of $10^{42}$ erg s$^{-1}$
319:            and km s$^{-1}$, respectively.}
320:  \enddata
321:  \end{deluxetable}
322: 
323: \placefigure{fig2}
324: 
325: \subsection{$\sigma$-$T$ relation}
326: 
327: The best fit $\sigma$-$T$ relation for clusters using 109 clusters
328: remains almost unchanged (Fig.3 and Table 4) as compared with 
329: the previous studies by  Wu, Fang \& Xu (1998; and references therein)
330: and Paper I: $\sigma\propto T^{0.65\pm0.03}$. Meanwhile, we have found
331: roughly the same relation for our  sample of 36 groups:
332: $\sigma\propto T^{0.64\pm0.08}$, which is also consistent with the previous
333: results within uncertainties (Ponman et al. 1996; Mulchaey \& Zabludoff 1998). 
334: For clusters, this relation alone indicates
335: that the intracluster gas may not be in isothermal and hydrostatic
336: equilibrium with the underlying gravitational potential of clusters.
337: However, the same conclusion is not strictly applicable to groups 
338: if the large uncertainty in  the presently fitted $\sigma$-$T$ relation 
339: is included.  Additionally, 
340: the ratios of specific energy in galaxies to that in gas for the 36
341: groups exhibit rather large scatters ranging from 0.3 to 3.5, with
342: an average value of $\beta_{spec}=0.85$. 
343: 
344: 
345: 
346:  \begin{deluxetable}{ccllll}
347:  \tablewidth{35pc}
348:  \scriptsize
349:  \tablecaption{The best fit $\sigma$-$T$ relations$^a$}
350:  \tablehead{
351:  \colhead{sample}& \colhead{number} &
352:  \colhead{OLS} & 
353:  \colhead{ODR} & 
354:  \colhead{$\tau$} &
355:  \colhead{$P$} }
356:  \startdata
357:  group   & 36  &    $\sigma=10^{2.51\pm0.19}T^{0.45\pm0.07}$ &
358:                     $\sigma=10^{2.53\pm0.01}T^{0.64\pm0.08}$ 
359:                &    $0.355$ &  $2.291\times10^{-3}$ \nl 
360: cluster  & 109 &    $\sigma=10^{2.53\pm0.03}T^{0.58\pm0.05}$ &
361:                     $\sigma=10^{2.49\pm0.02}T^{0.65\pm0.03}$
362:                &    $0.583$ &  $2.535\times10^{-19}$ \nl  
363: combined & 145 &    $\sigma=10^{2.53\pm0.19}T^{0.57\pm0.01}$ &
364:                     $\sigma=10^{2.51\pm0.01}T^{0.61\pm0.01}$
365:                &    $0.703$ &  $4.135\times10^{-36}$ \nl  
366: \tablenotetext{a}{$\sigma$ and $T$ are in units of km s$^{-1}$ and keV,
367: 	   respectively.}
368:  \enddata
369:  \end{deluxetable}
370: 
371: 
372: \placefigure{fig3}
373: 
374: 
375: 
376: \subsection{Co-consistency test}
377: 
378: The employment of the ODR fitting method essentially ensures that 
379: the best fit relations  between $L_x$, $T$ and $\sigma$ are self-consistent
380: (Paper I).  We now examine the co-consistency between 
381: these relations in the sense that these three relations are not independent.
382: Our strategy is as follows: We first derive the correlation between 
383: $\sigma$ and $T$ from the best fit $L_x$-$T$ and 
384: $L_x$-$\sigma$ relations listed in Table 2 and Table 3. We then
385: compare this `derived' $\sigma$-$T$ relation with the one 
386: obtained independently from our ODR fitting shown in Table 4. 
387: Our fitted relations should be acceptable if a good agreement between
388: the derived and fitted $\sigma$-$T$ relations is reached. Otherwise,
389: at least one of the three fitted relations will be questionable.
390: 
391: 
392: For cluster sample, combining the $L_x$-$T$ and $L_x$-$\sigma$ relations 
393: in Table 2 and Table 3 yields $\sigma\propto T^{0.53\pm0.04}$.
394: Therefore, at $2\sigma$ significance level the derived $\sigma$-$T$ relation 
395: is consistent with the directly fitted one $\sigma\propto T^{0.65\pm0.03}$.
396: As for the group sample,  the derived $\sigma$-$T$ relation reads
397: $\sigma\propto T^{2.37\pm0.90}$, which compares with 
398: the best fit one $\sigma\propto T^{0.64\pm0.08}$.
399: Regardless of the apparent disagreement between the mean slopes,  
400: the large $68\%$ confidence interval makes the two relations 
401: show no difference within $2\sigma$.
402: As a result, the three fitted relations for groups 
403: in Table 2 -- Table 4 are still consistent with each other 
404: when the $2\sigma$ uncertainties are taken into account.
405: Indeed, a visual examination of Fig.1--Fig.3 reveals that  
406: the data points of groups show very 
407: large dispersions especially on the $L_x$-$\sigma$ plane.
408: Either the sparse data points and/or the inclusion of a few unusual groups
409: in our fitting (e.g. HCG94, S49-142, etc.) may account for the 
410: the marginal co-consistency between the $L_x$-$T$, $L_x$-$\sigma$ and 
411: $\sigma$-$T$ relations for groups.
412: 
413: In Table 2 -- Table 3, we have also given the significance level 
414: for the resulting correlation coefficient for each data set, 
415: based on Kendall's $\tau$.  It appears that the probability of
416: erroneously rejecting the null hypothesis of no correlation
417: between $L_x$, $T$ and $\sigma$ is $P<0.2\%$ for all the cases. 
418: Therefore, it is unlikely that the correlations we have found 
419: for groups and clusters are an artifact of the small samples
420: and/or the statistical fluke.
421: 
422: 
423: 
424: \section{Discussion and conclusions}
425: 
426: 
427: \subsection{Global properties}
428: 
429: 
430: Groups and clusters constitute large and nearly virialized dynamical systems
431: in the Universe.  While the distribution of dark matter  in these systems
432: may exhibit a regularity such as the universal density profile,
433: the hot intragroup/intracluster gas could be affected by non-gravitational 
434: heating mechanisms (e.g. star formation), especially for poor
435: systems like groups of galaxies where even individual galaxies may 
436: make a nonnegligible contribution to the X-ray emission 
437: (Dell'Antonio et al.  1994; Mulchaey \& Zabludoff 1998).  
438: Therefore, the standard scenario that groups and clusters should be 
439: the similar dynamical systems at different mass ranges is only applicable 
440: to the distribution and evolution of the dark matter particles.
441: Whether or not the hot gas can be used as a good tracer of  the underlying
442: gravitational potential wells depends on how significant 
443: the X-ray emission of individual galaxies and the non-gravitational
444: heating would be, which in turn depends on how massive a dynamical system 
445: will be.  
446: Because groups of galaxies connect individual galaxies to clusters of 
447: galaxies, one may expect to detect the transition between 
448: ``galaxy-dominated'' and ``intracluster medium dominated'' properties 
449: occurring on group scales (e.g. Dell'Antonio et al.  1994). 
450: 
451: 
452: Such a scenario has essentially been justified by  
453: our analysis of the correlations between X-ray luminosity, temperature 
454: and velocity dispersion for  groups and clusters, 
455: although the current data especially for groups still have rather 
456: large uncertainties. It is remarkable that 
457: we have detected the similarity breaking in the $L_x$-$T$ and
458: $L_x$-$\sigma$ relations between massive systems (clusters) and low mass
459: ones (groups), indicating that the overall dominant X-ray emission and
460: heating mechanisms
461: are quite different in these two systems. Nevertheless, 
462: for the $\sigma$-$T$ relation we have not found an apparent discrepancy
463: between the two systems. Our result essentially agrees  
464: with the previous findings of Dell'Antonio et al. (1994), 
465: Ponman et al. (1996) and Mahdavi et al. (1997, 2000). 
466: 
467: 
468: 
469: \subsection{Uncertainties}
470: 
471: 
472: Major uncertainties in the presently fitted $L_x$-$T$, 
473: $L_x$-$\sigma$ and $\sigma$-$T$ relations for groups and clusters
474: come from the sparse (X-ray) data sets of poor clusters and groups.
475: Although an observational determination of these relations does not 
476: in principle require the completeness of the sample, the reliability and
477: significance of our fitting can be seriously affected by the small number
478: statistics. The marginal co-consistency between the three
479: relations for groups may be partially due to the small group sample.
480: It is generally believed that poor clusters and groups 
481: should all contain  hot X-ray emitting gas 
482: (e.g. Price et al. 1991; Ponman et al. 1996).
483: However, most of them can be hardly detected by current X-ray observations 
484: because of their low temperatures $T\sim0.1$--$1$ keV. Therefore, 
485: our group sample suffers from a selection bias, in which the present 
486: X-ray data are acquired by different authors with different criteria. 
487: For instance, about 1/3 of our group sample are HCGs, 
488: and the steepening of the $L_x$-$T$ relation for groups may have partially 
489: arisen from the contribution of these HCG populations. 
490: 
491: 
492: Our statistical results are sensitive to the linear fitting methods, especially
493: for the $L_x$-$T$ and $L_x$-$\sigma$ relations. This arises because 
494: the observed quantities, $L_x$, $T$ and $\sigma$, all have measurement
495: uncertainties, while OLS method ignores scatters around the horizontal
496: variable (e.g. $T$ or $\sigma$). At this point, the ODR fitting technique is 
497: strongly recommended.  However, not all the data points have information 
498: about their measurement uncertainties, or some error bars are difficult 
499: to evaluate. In this case, we have simplified the problem and employed 
500: the average values instead, in order to maximally use the available data 
501: points, which may have yielded further uncertainties in the ODR fitted 
502: relations. Note that the measurement uncertainties  in $L_x$ are 
503: relatively smaller than those in $T$ and $\sigma$ 
504: (see Fig.1 and Fig.2). This explains the 
505: significant difference in the resultant $L_x$-$T$ and $L_x$-$\sigma$ 
506: relations between  OLS and ODR methods.
507: 
508: 
509: \subsection{Physical implications}
510: 
511: Under the assumption that both gas and galaxies are in hydrostatic
512: equilibrium with the underlying gravitational potential of 
513: group/cluster, we would expect that the total X-ray luminosity via
514: thermal bremsstrahlung scales as (e.g. Paper I)
515: %1
516: \begin{equation}
517: L_x\propto T^{2.5}f_b^2 r_{c}^{-1} S_{gas},
518: \end{equation}
519: and 
520: %2
521: \begin{equation}
522: L_x\propto \sigma^{4} T^{1/2} f_b^2 r_{c}^{-1} S_{gal},
523: \end{equation}
524: where $f_b$ is the (gas) baryon fraction, $r_c$ is the core radius and 
525: $S$ is the so-called dimensionless structure factor which depends 
526: strongly on the power-index of gas/galaxy profile (e.g. the conventional
527: $\beta$ model or the King model) but weakly on the core radius.
528: Using the emission weighted temperature instead of $T$ in eq.(1) only 
529: leads to a modification to $S_{gas}$. Additionally, the velocity dispersion
530: of galaxies and the temperature of hot X-ray emitting gas satisfy
531: %3
532: \begin{equation}
533: \sigma \propto  T^{1/2}. 
534: \end{equation}
535: 
536: 
537: From our fitted $\sigma$-$T$ relations alone,  groups are still consistent with
538: $\sigma \propto  T^{1/2}$ within $2\sigma$ uncertainties, 
539: while a significant deviation from what is expected under 
540: the scenario of isothermal and hydrostatic equilibrium is found for clusters.
541: The latter is consistent with a number of similar studies on the issue 
542: (e.g. White, Jones \& Forman 1997; Wu et al. 1998; 1999).
543: Yet, it is unlikely that the intragroup gas is in the state of a perfect 
544: isothermal and hydrostatic equilibrium with the underlying gravitational
545: potential as allowed by the $\sigma$-$T$ relation within its $95\%$ 
546: significance interval.
547: This point can be further demonstrated by the apparent disagreement between 
548: the theoretical prediction of eq.(1), $L_x\propto T^2$, 
549: and our fitted $L_x$-$T$ relation  for groups in Table 2, 
550: $L_x\propto T^{5.57\pm1.79}$, unless the baryon fraction is 
551: assumed to vary with gas temperature. Meanwhile, the best fit $L_x$-$\sigma$
552: relation for groups, $L_x\propto \sigma^{2.35\pm0.21}$, differs from the 
553: theoretical expectation (eq.[2]) in conjunction with eq.(3): 
554: $L_x\propto \sigma^{4}$.  Taking these results as a whole, we feel that 
555: the currently available optical/X-ray data have already provided
556: convincing evidence for either the failure of isothermal and 
557: hydrostatic equilibrium hypothesis for intragroup gas or 
558: the significant contribution of X-ray emission from the individual 
559: group galaxies. Therefore, 
560: any cosmological applications of these fitted relations without considering
561: these effects could be misleading. 
562: 
563: 
564: The physical implications of the $L_x$-$T$, $L_x$-$\sigma$ and 
565: $\sigma$-$T$ relations for clusters have been extensively studied in
566: Paper I. For the group sample, there exists an apparent
567: difference between the theoretically predicted 
568: $L_x$-$T$ and $L_x$-$\sigma$ 
569: relations (eq.[1]) and the observationally determined ones. 
570: Dell'Antonio et al. (1994) attributed the flattening of $L_x$-$\sigma$ 
571: relation to the additional X-ray emission of individual group galaxies, while 
572: Ponman et al (1996) interpreted the steepening of the  $L_x$-$T$
573: relation as the result of the wind injection from galaxies.
574: The recent detection of the X-ray wakes in pool cluster A160 may give a 
575: strong support to these scenarios, i.e., a large fraction of 
576: X-ray emission in groups and poor clusters can be associated with
577: individual galaxies (Drake et al. 2000).
578: It thus deserves to  explore whether
579: this scenario can  quantitatively account 
580: for the reported $L_x$-$T$-$\sigma$ relations 
581: for groups. Alternatively, other mechanisms such as 
582: preheating by supernovae and substructure merging may
583: also contribute extra heating to the intragroup medium.
584: 
585: 
586: On the other hand, the prevailing determination of the baryon fraction of 
587: groups via hydrostatic equilibrium and thermal bremsstrahlung
588: could be seriously contaminated by the X-ray emission of individual 
589: group galaxies and other non-gravitational heating.
590: This might account for the relatively large variations of 
591: the presently derived baryon fractions  among different groups.
592: It is unlikely that a robust estimate of the baryon
593: fractions of groups and poor clusters within the framework of conventional 
594: dynamical models and their cosmic evolution can be achieved without
595: a better understanding of the various heating mechanisms.
596: Indeed, we cannot exclude the possibility that
597: some puzzles about the properties of the baryon fractions of groups
598: and clusters are due to the contamination of the X-ray emission of 
599: individual galaxies and the non-gravitational heating.  
600: For instance, the standard model predicts an increase 
601: in the baryon fraction with radius, and a universal value of $f_b$ 
602: at large radii cannot be reached unless the X-ray temperature is required
603: to rise in some clusters (Wu \& Xue 2000 and references therein).
604: The presence of such puzzle has essentially prevented the baryon fractions 
605: of groups and clusters from the cosmological applications.  
606: Inclusion of the contributions of other emission/heating sources  
607: in the estimate of the total
608: gravitational masses of groups and clusters may lead to a re-arrangement of
609: the radial distribution of baryon fraction.  Further work will thus be    
610: needed to explore whether these effects can resolve the above puzzle.
611: 
612: 
613: 
614: \acknowledgments
615: We thank the referee, Andisheh Mahdavi, for many valuable suggestions
616: and comments. This work was supported by 
617: the National Science Foundation of China, under Grant No. 19725311.
618: 
619: 
620: \clearpage
621: 
622: 
623: \begin{references}
624: \reference{}Beers, T. C., Kriessler, J. R., Bird, C. M., \& 
625: 		Huchra, J. P., 1995, \aj, 109, 874
626: \reference{}Buote, D. A., 1999, \mnras, 309, 685
627: \reference{}Burns, J. O., et al., 1996, \apj, 467, L49
628: \reference{}Cavaliere, A., Menci, N., \& Tozzi, P., 1997, \apj, 484, L21
629: \reference{}Cavaliere, A., Menci, N., \& Tozzi, P., 1999, \mnras, 308, 599
630: \reference{}David, L. P., Jones, C., \& Forman, W., 1995, \apj, 445, 578
631: \reference{}David, L. P., Jones, C., Forman, W., \& Daines, S.,
632: 				1994, \apj, 428, 544
633: \reference{}David, L. P., Slyz, A., Jones, C., Forman, W., \& Vrtilek, S. D., 
634: 	    	1993, \apj, 412, 479
635: \reference{}Davis, D. S., Mulchaey, J. S., \& Mushotzky, R. F.,
636: 		1999, \apj, 511, 34
637: \reference{}Dell'Antonio, I. P., Geller, M. J., \& Fabricant, D. G., 
638: 		1994, \aj, 107, 427
639: \reference{}Drake, N., Merrifield, M. R., Sakelliou, I., \& Pinkney, J. C.,
640: 		2000, \mnras, submitted (astro-ph/0002017)
641: \reference{}Fadda, D., Girardi, M., Giuricin, G., Mardirossian, F.,
642: 		\& Mezzetti, M., 1996, \apj, 473, 670
643: \reference{}Feigelson, E. D., \& Babu, G. J., 1992, \apj, 397, 55
644: \reference{}Fukugita, M., Hogan, C. J., \& Peebles, P. J. E., 1998, 
645: 		\apj, 503, 518
646: \reference{}Henry, J. P., et al., 1995, \apj, 449, 422
647: \reference{}Horner, D. J., Mushotzky, R. F., \& Scharf, C. A., 
648: 		1999, \apj, 520, 78
649: \reference{}Komossa, S., \& B\"ohringer, H., 1999, \aap, 344, 755
650: \reference{}Ledlow, M. J., Loken, C., Burns, J. O., Hill, J. M., \&
651: 				White, R. A., 1996, \aj, 112, 388
652: \reference{}Loewenstein, M., 1999, \apj, submitted (astro-ph/9910276)
653: \reference{}Mahdavi, A., B\"ohringer, H., Geller, M. J., \&
654: 				Ramella, M., 1997, \apj, 483, 68
655: \reference{}Mahdavi, A., B\"ohringer, H., Geller, M. J., \&
656: 				Ramella, M., 2000, \apj, in press
657: \reference{}Markevitch, M., Forman, W. R., Sarazin, C. L.,
658:                 \& Vikhlinin, A., 1998, \apj, 503, 77
659: \reference{}Mulchaey, J., Davis, D. S., Mushotzky, R. F., \&
660: 		Burstein, D., 1993, \apj, 404, L9
661: \reference{}Mulchaey, J. S., \& Zabludoff, A. I., 1998, \apj, 496, 73
662: \reference{}Navarro, J. F., Frenk, C. S., \& White, S. D. M., 1995, \mnras, 
663:             	275, 720 
664: \reference{}Pedersen, K., Yoshii, Y., \& Sommer-Larsen, J., 1997, \apj,
665: 		485, L17
666: \reference{}Pildis, R. A., Bregman, J. N., \& Evrard, A. E.,
667: 		1995, \apj, 443, 514
668: \reference{}Ponman, T. J., Bourner, P. D. J., Ebeling, H., \& B\"ohringer, H.,
669: 	    	1996, \mnras, 283, 690 
670: \reference{}Ponman, T. J., Cannon, D. B., \& Navarro, J. F., 1999, 
671:                \nat, 397, 135
672: \reference{}Price, R., Burns, J. O., Duric, N., \& Newberry, M. V., 
673: 		1991, \aj, 102, 14
674: \reference{}Quintana, H., \& Melnick, J., 1982, \aj, 87, 972
675: \reference{}Valageas, P., \& Schaeffer, R., 1999, \aap, in press 
676: 		(astro-ph/9909370)
677: \reference{}White, D. A.,  1999, \mnras, submitted (astro-ph/9909467)
678: \reference{}White, D. A., Jones, C., \& Forman, W., 1997, \mnras, 292, 419
679: \reference{}White, S. D. M., Navarro, J. F., Evrard, A. E., \& 
680:         	Frenk, C. S., 1993, \nat, 366, 429
681: \reference{}Wu, K. K. S., Fabian, A. C., \& Nulsen, P. E. J., 1999, 
682: 		\mnras, submitted (astro-ph/9907112)
683: \reference{}Wu, X.-P., Fang, L.-Z., \& Xu, W., 1998, \aap, 338, 813
684: \reference{}Wu, X.-P., \& Xue, Y.-J., 2000, \mnras, 311, 825
685: \reference{}Wu, X.-P., Xue, Y.-J., \& Fang, L.-Z, 1999, \apj, 524, 22 (Paper I)
686: \reference{}Zabludoff, A. I., \& Mulchaey, J. S., 1998, \apj, 496, 39
687: 
688: \end{references}
689: 
690: 
691: 
692: \clearpage
693: 
694: 
695: \figcaption{The $L_x$-$T$ relations for 184 clusters (open circles)
696: and 38 groups (filled circles). 
697: The dotted and solid lines are the best ODR fitted relations 
698: to the group and cluster samples, respectively.
699: \label{fig1}}
700: 
701: 
702: \figcaption{The $L_x$-$\sigma$ relations for 197 clusters 
703: and 59 groups. The notations are the same as in Fig.1.
704: \label{fig2}}
705: 
706: 
707: \figcaption{
708: The $\sigma$-$T$ relations for 109 clusters 
709: and 36 groups. The notations are the same as in Fig.1.
710: \label{fig3}}
711: 
712: \end{document}
713: 
714: 
715: 
716: 
717: 
718: 
719: 
720: 
721: 
722: 
723: 
724: 
725: 
726: 
727: 
728: 
729: 
730: