Within the $\sim$ 0.7 square degree imaged by ISOCAM, a total of 425 sources has been identified, among which 211 are seen at both 6.7 $\mu$ m and 14.3 $\mu$ m. The spatial distribution of the sources is shown in Fig. 1, where the "red'' sources [those with log₁₀( $\mbox{$F_\nu^{14.3}$ }/\mbox{$F_\nu^{6.7}$ }) > - 0.2$ , see below] are indicated as filled circles. These "red'' sources appear to be clustered into four main groupings: three sub-clusters in L1688, i.e., Oph A (West), Oph B (North-East), Oph EF (South) (see also Strom et al. 1995), as well as a new sub-cluster in L1689S.

Four bright embedded stars (S1, SR3, WL16, WL22) are spatially resolved by ISOCAM in both filters. Their extended mid-IR emission is most likely due to PAH-like molecules excited by relatively strong far-ultraviolet (FUV) radiation fields. These bright sources are displayed as open star symbols in Fig. 1 (SR3, S1, WL22, WL16 from right to left).

$\begin{figure} \par\includegraphics[width=13.5cm,clip]{ms9901f3.eps} \end{figure}$

Figure 3: a) Logarithmic ratio of the ISOCAM fluxes, log₁₀( $\mbox{$F_\nu^{14.3}$ }$ / $\mbox{$F_\nu^{6.7}$ }$ ), versus $\mbox{$F_\nu^{14.3}$ }$ displayed for the 207 ISOCAM sources seen in both filters (excluding the 4 early-type young stars). The purely ISOCAM color $\mbox{$F_\nu^{14.3}$ }$ / $\mbox{$F_\nu^{6.7}$ }$ is converted into an IR spectral index, $\mbox{$\alpha_\mathrm{IR}^{7-14}$ }$ , on the right axis, and can be used to identify sources with mid-IR excesses (above the dashed line). The completeness limit derived in Sect. 2.4 is displayed as a dotted line. b) Mid-IR color-color diagram expressed as a spectral index diagram $\mbox{$\alpha_\mathrm{IR}^{2-7}$ }$ versus $\mbox{$\alpha_\mathrm{IR}^{2-14}$ }$ for 175 ISOCAM sources also detected in the K band (m_6.7 and m_14.3on the right and upper axes are calculated as in Persi et al. 2000). The inclined heavy dashed line marks the limit to the right of which there is significant mid-IR excess as in a). The two vertical dashed lines mark the formal boundary between Class II, flat-spectrum, and Class I sources (see text for details). c) Classical near-IR color-color diagram, J-H versus H-K, for the 152 ISOCAM sources with JHK photometry.

A total of 89 previously classified YSOs lie within the area of the present survey: 2 Class 0, 69 Class I/II, and 18 Class III YSOs. ISOCAM detected 84 of these 89 YSOs (94%): 97% of the Class I/IIs, 94% of the Class IIIs, and none of the Class 0s. The two undetected Class I/IIs (GY256, GY257), and the undetected Class III (IRS50), are located very close ( $\sim$ 10 $^{\prime\prime}$ -15 $^{\prime\prime}$ ) to bright sources (WL6 and IRS48, respectively), which may account for their non-detection. The two Class 0 objects (VLA 1623 and IRAS 16293-2422) are deeply embedded within massive, cold circumstellar envelopes which are probably opaque at 6.7 $\mu$ m and 14.3 $\mu$ m (see André et al. 1993) and too weak to be detected by ISOCAM.

3.2 A population of new YSOs with mid-IR excess

As can be seen in Fig. 3, the mid-IR regime is ideal to detect and characterize the excess emission due to circumstellar disks around young stars. In Fig. 3a, the 207 sources detected at 6.7 $\mu$ m and 14.3 $\mu$ m (excluding S1, SR3, WL16, WL22) are shown in a diagram which displays the logarithmic flux ratio log₁₀( $\mbox{$F_\nu^{14.3}$ }/\mbox{$F_\nu^{6.7}$ }$ ) against the mid-IR flux $\mbox{$F_\nu^{14.3}$ }$ . On the right axis of the diagram, the flux ratio $\mbox{$F_\nu^{14.3}$ }/\mbox{$F_\nu^{6.7}$ }$ has been converted into a classical IR spectral index, $\mbox{$\alpha_\mathrm{IR}$ }=$ d $\log_{10}(\lambda F_\lambda)/$ d $\log_{10}(\lambda)$ (e.g. WLY89), calculated between 6.7 $\mu$ m and 14.3 $\mu$ m, i.e., $\mbox{$\alpha_\mathrm{IR}^{7-14}$ }$ . In this diagram, two groups of sources can clearly be distinguished. The lower ratio group has log₁₀( $\mbox{$F_\nu^{14.3}$ }/\mbox{$F_\nu^{6.7}$ }) \sim -0.7$ , i.e., a spectral index on the order of $\mbox{$\alpha_\mathrm{IR}^{7-14}$ }\sim -3.0$ , which is the value expected for simple photospheric blackbody emission in the Rayleigh-Jeans regime. The dispersion around $\mbox{$\alpha_\mathrm{IR}^{7-14}$ }\sim -3.0$ is obviously larger for weaker sources. This is mainly due to increasing photometric uncertainty with decreasing flux. The higher ratio group consists of "red'' sources defined by log₁₀( $\mbox{$F_\nu^{14.3}$ }/\mbox{$F_\nu^{6.7}$ }$ ) > -0.2 (cf. Nordh et al. 1996). Most of these actually have $-0.2 \leq$ log₁₀( $\mbox{$F_\nu^{14.3}$ }/\mbox{$F_\nu^{6.7}$ }) \leq 0.3$ , i.e., $-1.6\, \lower.5ex\hbox{$\buildrel < \over \sim$ }\, \mbox{$\alpha_\mathrm{IR}^{7-14}$ }\, \lower.5ex\hbox{$\buildrel < \over \sim$ }\, 0$ , which is typical of classical T Tauri stars (hereafter CTTS). This range of $\mbox{$\alpha_\mathrm{IR}^{7-14}$ }$ roughly delineates the domain of Class II YSOs (e.g. Adams et al. 1987, AM94; Greene et al. 1994), and is usually interpreted in terms of optically thick circumstellar disk emission (e.g. Lada & Adams 1992). In Fig. 3, the new ISOCAM sources are distinguished from the previously known cluster members by different symbols. One can see in Fig. 3a that all the previously known Class II sources but one lie above the dividing line for red sources, while all the previously known Class III sources but two lie below it.

**Table 2:** Class I YSOs.
ISO	Identification	$\mbox{$\alpha_\mathrm{IR}^{2-14}$ }$	$\mbox{$L_\mathrm{bol}$ }$ ^b
#^a			[ $L_\odot$ ]
182	IRS54/GY378	1.76	6.6
143	IRS44/GY269	1.57	8.7
137	CRBR85	1.48	0.36
99	LFAM26/GY197	1.25	0.064
29	GSS30/GY6	1.20	21.
31	LFAM1	1.08	0.13
65	WL12/GY111	1.04	2.6
108	EL29/GY214	0.98	26.
141	IRS43/GY265	0.98	6.7
145	IRS46/GY274	0.94	0.62
21	CRBR12	0.91	0.42
209	IRS67/L1689-IRS6	0.74	1.5
54	GY91/CRBR42	0.70	0.17
134	WL6/GY254	0.59	1.7
159	IRS48/GY304^c	0.18	7.4
167	IRS51/GY315^c	-0.04	1.1

^a The ISO number refers to the numbering of Table 1 (available only in electronic form at http://cdsweb.u-strasbg.fr/).
^b $\mbox{$L_\mathrm{bol}$ }$ is estimated by integrating under the observed SED from 1.2 $\,\mu$ m to 60 $\,\mu$ m or 100 $\,\mu$ m and extrapolating from 60 $\,\mu$ m or 100 $\,\mu$ m to 200 $\,\mu$ m with a spectral index of $\mbox{$\alpha_\mathrm{IR}$ }=-1.0$ . For the 9 weak sources (ISO $\,$ # 137, 99, 31, 65, 146, 21, 209, 54 and 133) without reliable IRAS fluxes, $\mbox{$L_\mathrm{bol}$ }$ is estimated using a typical $\mbox{$L_\mathrm{bol}$ }/\mbox{$L_\mathrm{cal}$ }(6.7{-}14.3~\mu$ m) ratio of 9.8 (see Sect. 4.3).
^c Although ISO159 $\,=\,$ IRS48 and ISO167 $\,=\,$ IRS51 formally lie below our practical Class I-Class II limit ( $\mbox{$\alpha_\mathrm{IR}^{2-14}$ }= 0.55$ ), they are still considered as Class I YSOs here (cf. WLY89 and AM94).

The object WL19, classified as a Class I YSO by WLY89 and as a reddened Class II by AM94 (see also Lada & Wilking 1984), is here found to be a "blue'' source in the mid-IR range ( $\mbox{$\alpha_\mathrm{IR}^{7-14}$ }=-1.8\pm0.2$ ). This object may correspond to a luminous Class III star located behind the cloud (see also Comerón et al. 1993). Although GY12 formally has a "red'' mid-IR spectral index ( $\mbox{$\alpha_\mathrm{IR}^{7-14}$ }=-0.5\pm0.5$ ) here, we still consider it as a Class III object (cf. Greene et al. 1994). (The mid-IR color is highly uncertain since GY12 is only marginaly resolved at 14.3 $\mu$ m from its bright Class I neighbor GSS30.) Finally the Class III source DoAr21 has a borderline mid-IR spectral index ( $\mbox{$\alpha_\mathrm{IR}^{7-14}$ }=-1.5\pm0.1$ ) but is kept as a Class III object since its color between 2.2 $\mu$ m and 14.3 $\mu$ m corresponds to $\mbox{$\alpha_\mathrm{IR}^{2-14}$ }=-2.0\pm0.1$ .

A total of 71 sources are identified for the first time as mid-IR excess objects in Fig. 3a. These new "red'' sources are most likely all embedded YSOs, i.e., members of the $\mbox{$\rho$ ~Ophiuchi}$ cluster. Since dust extinction is roughly the same at 6.7 $\mu$ m and 14.3 $\mu$ m (Rieke & lebofsky 1985; Lutz 1999), any background source should be intrinsically red in order to contaminate the sample of red YSOs. Based on the Galactic model by Wainscoat et al. (1992), the vast majority of background objects should appear "blue'' (cf. Fig. 2). Seven background giant stars (GY45, GY65, VSSG6, GY232, GY351, GY411, and GY453 - cf. Luhman & Rieke 1999), and a known foreground dwarf (HD 148352 - Garrison 1967) are detected, which all have blue mid-IR colors ( $\mbox{$\alpha_\mathrm{IR}^{7-14}$ }= -2.7, -2.0, -3.1, -3.3, -2.9, -3.2, -3.1$ , and -3.1, respectively).

**Table 3:** Class II YSOs.
ISO	Identification	$\mbox{$\alpha_\mathrm{IR}^{2-14}$ }$	M_J	M_H	$\mbox{$A_V$ }$	$\mbox{$L_\star$ }$	$\mbox{$L_\mathrm{disk}$ }$
#^a			[mag]	[mag]	[mag]	[ $L_\odot$ ]	[ $L_\odot$ ]
170	B162741-244645	0.51	-	6.1	24.9	0.018	0.047
103	WL17/GY205	0.42	5.2	-	22.5	0.12	0.76
124	IRS37/GY244	0.35	-	2.4	36.9	0.99	0.50
112	GY224	0.34	-	2.4	36.2	1.1	0.56
118	IRS33/GY236	0.32	-	3.6	38.3	0.28	0.15
33	GY11	0.31	10.1	-	2.7	0.001	0.010
119	IRS35/GY238	0.30	-	3.4	45.7	0.34	0.11
129	WL3/GY249	0.23	-	2.2	42.2	1.3	0.34
75	GY144	0.20	-	5.9	26.8	0.023	0.030
147	IRS47/GY279	0.17	2.6	-	26.8	1.9	1.8
46	VSSG27/GY51	0.17	5.2	-	21.6	0.11	0.30
127	GY245	0.17	6.7^b	-	24.7	0.023	0.11
161	GY301	0.12	-	2.1	44.9	1.5	0.25
132	IRS42/GY252	0.08	2.1	-	27.7	3.1	2.5
77	GY152	0.05	-	-	-	0.037^c	0.015
70	WL2/GY128	0.05	-	2.3	38.6	1.1	0.27
165	GY312	0.03	6.6	-	14.8	0.027	0.064
85	CRBR51	0.03	-	-	-	0.025^c	0.010
175	GY344	0.02	6.5	-	17.4	0.030	0.074
26	CRBR15	0.01	6.7	-	14.6	0.022	0.061
139	GY260	-0.03	-	3.7	40.3	0.24	0.058
37	LFAM3/GY21	-0.06	4.5	-	14.5	0.25	0.33
121	WL20/GY240	-0.07	3.3	-	16.5	0.86	0.67
51	B162636-241554	-0.09	4.4	-	7.6	0.28	0.43
95	WL1/GY192	-0.11	4.9	-	20.9	0.15	0.16
122	IRS36/GY241	-0.11	-	4.9	34.3	0.066	0.025
171	GY323	-0.12	-	4.0	30.9	0.19	0.070
107	GY213	-0.15	5.6	-	19.4	0.075	0.089
76	GY146	-0.16	-	4.1	43.5	0.17	0.025
120	IRS34/GY239	-0.23	-	2.4	34.8	1.1	0.19
144	IRS45/GY273	-0.24	3.6	-	19.5	0.66	0.45
204	L1689-IRS5^d	-0.25	2.6	-	12.8	1.9	1.4
17	GSS26	-0.30	3.2	-	22.6	0.95	0.43
93	GY188	-0.36	6.3	-	20.5	0.034	0.020
98	GY195	-0.36	5.5	-	20.6	0.087	0.063
53	GY84	-0.39	5.8	-	14.1	0.060	0.037
190	GY450	-0.39	9.2	-	8.6	0.002	0.008
23	SKS1-10	-0.41	7.2	-	8.2	0.013	0.018
13	B162607-242725	-0.41	4.6	-	19.5	0.22	0.12
117	GY235	-0.43	4.9	-	9.9	0.16	0.12
212	L1689-IRS7^d	-0.44	2.6	-	14.7	1.8	0.64
79	GY154	-0.44	-	5.5	24.3	0.033	0.013
24	VSSG1	-0.49	3.2	-	17.1	0.97	0.49
3	IRS3	-0.50	4.3	-	4.6	0.31	0.28
39	S2/GY23	-0.51	2.0	-	11.8	3.7	1.6
140	GY262	-0.52	3.5	-	23.7	0.71	0.20
40	EL24	-0.54	1.8	-	10.0	4.5	2.1
52	VSSG4/GY81	-0.54	5.2	-	17.6	0.11	0.058
154	GY291	-0.60	4.3	-	23.0	0.29	0.058
94	B162703-242007	-0.61	-	6.6	16.6	0.010	0.006
128	WL4/GY247	-0.67	3.2	-	19.5	1.1	0.24
59	WL7/GY98	-0.69	4.1	-	27.2	0.37	0.061
84	WL21/GY164	-0.70	7.8	-	14.2	0.007	0.008
67	GSS39/GY116	-0.72	3.1	-	16.3	1.1	0.25
41	GY29	-0.73	5.1	-	19.1	0.13	0.043
88	SR24N/GY168	-0.74	2.5	-	7.7	2.1	0.76
35	GY15	-0.75	6.2	-	11.5	0.040	0.019
164	GY310	-0.75	6.4	-	4.0	0.032	0.024
63	GY109	-0.77	5.8	-	14.6	0.064	0.022
151	GY284	-0.77	4.9	-	7.6	0.15	0.065
43	GY33	-0.78	4.2	-	15.5	0.36	0.092
138	B162726-241925	-0.78	7.9	-	11.2	0.006	0.004
36	GSS31/GY20	-0.79	1.5	-	6.1	5.9	1.6
177	GY352	-0.79	5.2	-	17.7	0.11	0.031
197	B162821-244246	-0.81	6.0	-	20.3	0.050	0.009
110	SR21/VSSG23	-0.81	1.9	-	3.5	4.0	1.7
166	GY314	-0.83	3.4	-	6.4	0.80	0.26
9	SKS1-4	-0.85	6.0	-	9.5	0.051	0.022
12	B162604-241753	-0.86	6.6	-	13.4	0.027	0.009
155	GY292	-0.90	2.7	-	10.8	1.6	0.37

**Table 3:** continued.
ISO	Identification	$\mbox{$\alpha_\mathrm{IR}^{2-14}$ }$	M_J	M_H	$\mbox{$A_V$ }$	$\mbox{$L_\star$ }$	$\mbox{$L_\mathrm{disk}$ }$
#^a			[mag]	[mag]	[mag]	[ $L_\odot$ ]	[ $L_\odot$ ]
88	SR24S/GY167	-0.91	2.5	-	5.9	2.2	0.72
19	GSS29	-0.91	2.9	-	9.4	1.4	0.28
115	WL11/GY229	-0.92	6.3	-	13.8	0.037	0.015
196	WSB60^e	-0.92	4.9	-	2.8	0.16	0.075
30	GY5	-0.92	6.3	-	2.5	0.036	0.019
176	GY350	-0.94	6.4	-	6.4	0.033	0.014
72	WL18/GY129	-0.94	4.7	-	10.4	0.19	0.061
163	IRS49/GY309	-0.96	3.1	-	10.1	1.1	0.23
193	B162812-241138	-0.99	6.2	-	6.2	0.039	0.016
2	B162538-242238	-1.00	4.6	-	10.9	0.22	0.063
78	VSSG5/GY153	-1.02	3.8	-	19.7	0.55	0.056
86	IRS26/GY171	-1.04	5.2	-	19.9	0.12	0.015
105	WL10/GY211	-1.05	3.4	-	12.5	0.78	0.12
87	B162658-241836	-1.06	6.0	-	14.2	0.050	0.011
160	B162737-241756	-1.08	7.2	-	4.5	0.014	0.006
83	B162656-241353	-1.08	4.2	-	11.1	0.33	0.058
32	GY3	-1.09	6.5	-	0.4	0.029	0.015
1	IRS2	-1.09	3.4	-	4.9	0.79	0.15
185	GY397	-1.10	6.1	-	4.1	0.045	0.016
6	SR4/IRS12	-1.12	2.8	-	1.9	1.5	0.37
142	VSSG25/GY267	-1.13	4.0	-	9.8	0.42	0.065
20	DoAr24/GSS28	-1.14	3.5	-	1.8	0.72	0.16
178	GY371	-1.16	5.2	-	7.1	0.11	0.026
62	GSS37/GY110	-1.19	2.7	-	8.5	1.6	0.18
89	WL14/GY172	-1.24	5.8	-	17.2	0.059	0.006
194	B162813-243249	-1.24	5.2	-	5.9	0.12	0.024
116	B162713-241818	-1.27	3.7	-	10.7	0.57	0.046
106	B162708-241204	-1.33	4.5	-	8.0	0.24	0.027
102	GY204	-1.33	6.2	-	1.5	0.039	0.011
168	SR9/IRS52	-1.35	2.7	-	-0.7	1.6	0.23
199	SR13	-1.37	3.4	-	-0.9	0.78	0.12
187	SR10/GY400	-1.39	4.1	-	-3.2	0.38	0.052
68	VSS27	-1.53	2.6	-	5.6	1.9	0.13
172	GY326	-1.56	5.3	-	8.5	0.11	0.008
56	WSB37/GY93	-1.58	4.7	-	1.2	0.20	0.024
38	DoAr25/GY17	-1.58	3.4	-	0.7	0.83	0.11
-	GY256	-	-	4.5	34.5	0.11	-
-	GY257	-	-	4.1	33.2	0.17	-
90	WL22/GY174	-	-	-	-	29.^f	-
92	WL16/GY182	-	-	-	-	44.^f	-
207	IRAS16289-2457	-	-	-	-	1.3^c	0.51
123	New^g	-	-	-	-	0.077^c	0.031
150	New^g	-	-	-	-	0.21^c	0.083
195	New^g	-	-	-	-	0.14^c	0.058
200	New^g	-	-	-	-	0.95^c	0.38
201	New^g	-	-	-	-	0.025^c	0.010
202	New^g	-	-	-	-	0.024^c	0.010
203	New^g	-	-	-	-	0.71^c	0.28
205	New^g	-	-	-	-	0.17^c	0.067
206	New^g	-	-	-	-	0.099^c	0.040
208	New^g	-	-	-	-	0.058^c	0.023
210	New^g	-	-	-	-	0.067^c	0.027
211	New^g	-	-	-	-	0.068^c	0.027

The sample of 21 Class II sources with $-0.05 \leq \mbox{$\alpha_\mathrm{IR}^{2-14}$ }\leq 0.55$ might contain a significant population of transition objects (flat-spectrum objects) between Class I protostars and Class II T Tauri stars.
^a The ISO number refers to the numbering in Table 1 (available only in electronic form at http://cdsweb.u-strasbg.fr/).
^b The J-band flux of ISO127 $\,=\,$ GY245 is taken from Greene et al. (1994).
^c ISO77 $\,=\,$ GY152, ISO85 $\,=\,$ CRBR51, and the last 13 sources are not detected in all near-IR bands, and no reliable M_J or M_H can be estimated. $\mbox{$L_\star$ }$ has been obtained using Eq. (8) (Sect. 4.2).
^d ISO204 $\,=\,$ L1689-IRS5 and ISO212 $\,=\,$ L1689-IRS7 refer to the IR sources listed by Greene et al. (1994) in L1689.
^e ISO196 $\,=\,$ WSB60 corresponds to the source B162816-243657 in Barsony et al. (1997).
^f Since ISO90 $\,=\,$ WL22 and ISO92 $\,=\,$ WL16 are young early-type stars (see Sect. 3.1), $\mbox{$L_\star$ }$ cannot be derived accurately using the method described in Sect. 4.1. The quoted luminosities for WL22 and WL16 are taken from WLY89 and Comerón et al. (1993), respectively. They have been scaled to $d=140\,$ pc.
^g The adopted names and the J2000 coordinates of the completely new IR sources are given in Table 1.

The newly identified cloud members nicely extend the previously known Class I/II population toward low IR fluxes. While previous studies could only identify sources with $\mbox{$F_\nu^{14.3}$ }\lower.5ex\hbox{$\buildrel > \over \sim$ }100$ mJy, the present census is complete for objects down to $\mbox{$F_\nu^{14.3}$ }\sim 15$ mJy. Altogether, a sample of 139 Class I/Class II YSOs is identified, of which 71 are new members. The present survey has thus allowed us to more than double the number of recognized YSOs with circumstellar IR excess in the $\mbox{$\rho$ ~Ophiuchi}$ cloud.

The 139 red YSOs are listed in Table 2 (Class I YSOs) and Table 3 (Class II YSOs) by decreasing order of $\mbox{$\alpha_\mathrm{IR}^{2-14}$ }$ . Twelve of them (last entries of Table 3) are completely new sources with respect to published IR surveys.

3.3 Class I versus Class II YSOs

In Fig. 3, the YSOs classified as Class I and Class II by AM94 and Greene et al. (1994) are shown as filled stars and filled circles, respectively. In the log₁₀( $\mbox{$F_\nu^{14.3}$ }$ / $\mbox{$F_\nu^{6.7}$ }$ ) versus $\mbox{$F_\nu^{14.3}$ }$ diagram, there is no clear color gap between the two classes of objects, even though the Class I YSOs tend to lie in the upper part of the "red'' group (see Fig. 3a). Since extinction has a negligible effect on the mid-IR ratio $\mbox{$F_\nu^{14.3}$ }$ / $\mbox{$F_\nu^{6.7}$ }$ , this suggests that Class I and Class II objects have fairly similar intrinsic colors between 6.7 $\mu$ m and 14.3 $\mu$ m.

The classical IR spectral index calculated from 2 $\mu$ m to 10 $\mu$ m (or 25 $\mu$ m) (e.g. Lada & Wilking 1984 and WLY89) appears to provide a better way of discriminating between envelope-dominated Class I YSOs and disk-dominated Class II sources (see Fig. 3b). In particular, millimeter continuum mapping of optically thin circumstellar dust emission confirms that, apart from a few important exceptions (e.g., WL22, WL16, WL17, IRS37, IRS47), the $\mbox{$\rho$ ~Ophiuchi}$ objects selected on the basis of $\mbox{$\alpha_\mathrm{IR}^{2-10}$ }\lower.5ex\hbox{$\buildrel > \over \sim$ }0$ are indeed Class I protostars surrounded by spheroidal envelopes (AM94, MAN98).

As expected, the previously known Class I YSOs are well concentrated in the upper-right part of the $\mbox{$\alpha_\mathrm{IR}^{2-7}$ }$ versus $\mbox{$\alpha_\mathrm{IR}^{2-14}$ }$ diagram of Fig. 3b (where $\mbox{$\alpha_\mathrm{IR}^{2-7}$ }$ and $\mbox{$\alpha_\mathrm{IR}^{2-14}$ }$ are close to the $\mbox{$\alpha_\mathrm{IR}^{2-10}$ }$ index used in previous studies - e.g., AM94 and Greene et al. 1994). There is also a hint of two gaps in this diagram at $\mbox{$\alpha_\mathrm{IR}^{2-14}$ }\sim 0.5$ and $\mbox{$\alpha_\mathrm{IR}^{2-14}$ }\sim 0.0$ , which roughly bracket the regime of flat-spectrum sources as defined by Greene et al. (1994). These may represent a distinct population of transition objects between Class I and Class II (e.g. Calvet et al. 1994).

Here, we thus consider sources with $\mbox{$\alpha_\mathrm{IR}^{2-14}$ }> 0.55$ as Class I YSOs (Table 2), sources with $-0.05 < \mbox{$\alpha_\mathrm{IR}^{2-14}$ }< 0.55$ as candidate flat-spectrum objects (see Table 3), and sources with $\mbox{$\alpha_\mathrm{IR}^{2-14}$ }< 0.55$ as Class II YSOs (Table 3). These limiting indices are displayed in Fig. 3b. In the following, the candidate flat-spectrum sources will be treated as Class II YSOs.

3.4 Mid-IR excess versus near-IR excess

Most of the ISOCAM sources (e.g. 90% of the Class II sources) were also detected in the near-IR (JHK) survey of Barsony et al. (1997). Comparison between Figs. 3b and 3c illustrates the advantage of mid-IR measurements for selecting sources with intrinsic circumstellar IR excesses. While the red and blue groups of Fig. 3a are well separated in the mid-IR diagram of Fig. 3b, they blend together in the near-IR diagram of Fig. 3c.

Figure 3c also shows that most of the ISOCAM-selected YSOs lie within the reddening band associated with the intrinsic locus of CTTSs as derived by Meyer et al. (1997) in Taurus. The few exceptions, which lie to the right of the reddening band, correspond to Class I and flat-spectrum YSOs.

3.5 The Class III YSO census in $\mbox{$\rho$ ~Ophiuchi}$

While Class I/II YSOs are easily recognized in the mid-IR range thanks to their strong IR excesses, Class III objects are difficult to identify without deep X-ray and/or radio centimeter continuum observations. We have used the ROSAT X-ray surveys of Casanova et al. (1995) and Grosso et al. (2000), along with the VLA radio surveys by, e.g., André et al. (1987) and Stine et al. (1988) to build up a sample of bona-fide Class III YSOs covered by the present survey. With the additional Class III candidate WL19 (see Sect. 3.2 above), there are 38 such YSOs which are listed in order of decreasing $\mbox{$\alpha_\mathrm{IR}^{2-14}$ }$ in Table 4.

**Table 4:** Class III YSOs.
ISO	Identification	$\mbox{$\alpha_\mathrm{IR}^{2-14}$ }$	M_J	M_H	$\mbox{$A_V$ }$	$\mbox{$L_\star$ }$
#^a		( $\mbox{$\alpha_\mathrm{IR}^{2-7}$ }$ )	[mag]	[mag]	[mag]	[ $L_\odot$ ]
114	WL19/GY227^b	-0.05	-	-1.0	73.4	52.
125	WL5/GY246^b	-1.02	-	-0.7	59.8	39.
34	GY12^b	-1.06	3.3	-	19.5	0.89
58	WL8/GY96^b	-1.11	1.9	-	35.5	4.2
152	GY289^b	-1.33	3.3	-	27.5	0.90
198	SR20	-1.65	1.9	-	4.0	4.0
133	GY253^b	-1.73	3.3	-	31.2	0.85
45	LFAM8/SKS1-19^b	-1.82	4.2	-	25.2	0.34
80	GY156^b	-2.00	3.4	-	22.5	0.84
10	DoAr21/GSS23^b	-2.00	0.7	-	6.0	15.
27	WSB28	-2.06	4.1	-	4.3	0.36
149	B162730-244726	-2.20	3.7	-	10.3	0.61
135	VSSG22	-2.23	3.0	-	17.1	1.2
64	VSSG11^b	-2.24	3.4	-	15.2	0.83
73	VSSG3/GY135	-2.26	2.3	-	15.7	2.8
180	VSSG14/GY372	-2.51	2.2	-	5.5	3.0
184	IRS55/GY380	-2.65	2.6	-	6.0	1.8
11	VSSG19^b	-2.75	3.8	-	3.9	0.50
7	GSS20	-2.84	3.1	-	4.7	1.1
60	GY101^b	( 0.00)	-	2.0	55.5	2.0
61	GY103^b	(-0.45)	-	2.6	48.5	1.1
126	GY248^b	(-0.82)	4.2	-	25.5	0.36
101	IRS30/GY203^b	(-0.99)	-	2.2	36.5	1.6
14	B162607-242742	(-1.43)	3.5	-	20.6	0.70
8	B162601-242945	(-1.96)	3.9	-	8.1	0.47
183	GY377^b	(-1.98)	4.0	-	16.0	0.40
157	GY296^b	(-2.02)	5.5	-	5.1	0.081
4	B162541-242138	(-2.05)	5.8	-	6.7	0.060
69	GY122	(-2.17)	5.5	-	2.7	0.080
96	GY193	(-2.24)	4.2	-	7.4	0.34
97	GY194	(-2.35)	4.1	-	9.1	0.40
188	GY410	(-2.36)	4.0	-	10.2	0.43
66	GY112	(-2.42)	4.2	-	3.6	0.35
130	SR12/GY250^b	(-2.48)	3.4	-	1.2	0.83
5	IRS10	(-2.78)	2.7	-	5.4	1.7
-	IRS50/GY306	-	3.7	-	11.5	0.60
16	SR3/GSS25	-	-	-	-	100.^c
48	S1/GY70^b	-	-	-	-	1100.^c

^a The ISO number refers to the numbering in Table 1 (available only in electronic form at http://cdsweb.u-strasbg.fr/).
^b Class III YSOs located inside the CS contours of Fig. 1 (see Sects. 3.5 and 4.4).
^c For the two B stars ISO16 $\,=\,$ SR3 and ISO48 $\,=\,$ S1 (e.g. Elias 1978), $\mbox{$L_\star$ }$ cannot be derived using the method described in Sect. 4.1. The quoted values are taken from Lada & Wilking (1984), and have been scaled to $d=140\,$ pc.

This Class III sample is unfortunately not as complete as the Class I and Class II samples discussed above. According to Grosso et al. (2000), the number of Class IIIs may be roughly as large as the number of Class IIs: above their typical X-ray detection limit of $L_{\rm X} \sim 3\times 10^{29}$ ergs^-1 (corresponding to $L_\star \sim 0.3\, L_\odot$ - see Fig. 7 of Grosso et al.), they found a Class III/Class II number ratio of 19/22 in the ROSAT-HRI/ISO-ISOCAM overlapping survey area. If this ratio is representative of the complete population of young stars in $\mbox{$\rho$ ~Ophiuchi}$ , the total number of Class IIs found here (123 objects) suggests that as many as 106 Class IIIs may be present in the cluster down to $\mbox{$L_\star$ }\sim 0.03\,L_\odot$ (our completeness level for Class IIs, see Sect. 4.4). A total of 38 Class IIIs are already known within the ISOCAM survey area, so that $\sim$ 68 unknown Class IIIs may remain to be found. Since it was noted in Sect. 2.4 that $\sim$ 30 sources detected at 6.7 $\mu$ m might be unidentified cluster members, about half of the missing Class IIIs may have been actually seen by ISOCAM.

We also note that a large proportion (80/123) of the Class II sources are closely associated with the densest part of L1688 (see Fig. 1). Assuming the same proportion applies to Class IIIs, we would expect $\sim$ 44 unknown Class III sources to be located within the CS contours of Fig. 1. A total of 39 unclassified ISOCAM sources (also detected by Barsony et al. 1997) lie within these CS contours where the number of detected background stars should be small due to high cloud extinction. Most of these 39 sources might thus be yet unidentified Class III YSOs. These candidate Class III sources are listed in order of decreasing $\mbox{$\alpha_\mathrm{IR}^{2-14}$ }$ in Table 5.

**Table 5:** Class III candidates located within the CS contours of Fig. 1.
ISO	Identification	$\mbox{$\alpha_\mathrm{IR}^{2-14}$ }$	M_J	M_H	$\mbox{$A_V$ }$	$\mbox{$L_\star$ }$
#^a		( $\mbox{$\alpha_\mathrm{IR}^{2-7}$ }$ )	[mag]	[mag]	[mag]	[ $L_\odot$ ]
191	GY463	-1.46	4.4	-	25.2	0.28
44	B162628-241543	-1.63	3.9	-	21.7	0.45
81	VSSG7/GY157	-2.13	2.1	-	29.8	3.4
91	VSSG8/GY181	-2.20	2.4	-	22.9	2.4
28	B162621-241544	-2.23	2.9	-	15.5	1.4
18	SKS1-7	-2.38	3.6	-	17.4	0.64
25	CRBR17	( 0.68)	8.3	-	10.1	0.004
153	GY290	( 0.41)	-	3.7	41.5	0.31
100	B162705-244013^b	( 0.12)	-	-	-	-
109	GY215^b	( 0.05)	-	-	-	-
104	GY207	(-0.31)	-	4.6	39.5	0.11
49	B162636-241811^b	(-0.36)	-	-	-	-
15	CRBR4	(-0.41)	-	1.9	58.9	2.2
146	GY278	(-0.58)	-	2.4	47.8	1.3
136	GY258	(-0.72)	8.2	-	10.8	0.004
55	IRS16/GY92	(-0.80)	-	5.0	29.1	0.079
173	IRS53/GY334	(-0.81)	-	2.0	40.2	2.1
71	GY130	(-0.87)	-	3.7	38.8	0.31
162	GY309	(-1.01)	-	3.3	42.5	0.47
50	B162636-241902	(-1.06)	6.6	-	14.0	0.025
57	B162641-241801	(-1.22)	-	3.1	32.3	0.58
111	WL9/GY220	(-1.27)	5.5	-	21.6	0.086
22	B162618-241712	(-1.50)	4.2	-	26.8	0.34
113	IRS32/GY228	(-1.52)	3.6	-	18.5	0.62
82	GY163	(-1.53)	-	4.1	32.8	0.19
131	GY255	(-1.57)	4.2	-	21.6	0.35
169	GY322	(-1.59)	4.0	-	16.3	0.43
47	IRS14/GY54	(-1.85)	4.7	-	16.6	0.21
189	GY412	(-1.90)	5.2	-	17.6	0.12
181	GY373	(-1.91)	6.4	-	7.7	0.032
174	GY346	(-1.93)	5.6	-	19.5	0.075
74	IRS20/GY143	(-1.97)	3.8	-	16.0	0.50
179	GY370	(-1.99)	5.4	-	10.0	0.094
148	GY283	(-2.09)	4.3	-	13.1	0.32
42	VSSG29/GY37	(-2.14)	6.8	-	6.5	0.021
192	GY472	(-2.14)	5.9	-	20.1	0.054
186	GY398	(-2.20)	5.2	-	5.2	0.11
158	GY297	(-2.22)	6.0	-	0.7	0.047
156	GY295	(-2.52)	4.3	-	4.8	0.32

^a: The ISO number refers to the numbering in Table 1 (available only in electronic form at http://cdsweb.u-strasbg.fr/).
^b: $\mbox{$A_V$ }$ and $\mbox{$L_\star$ }$ were not derived for these sources as they were detected only in the K band by Barsony et al. (1997).

3 Identification and nature of the mid-IR sources

3.1 Statistics of detections

3.2 A population of new YSOs with mid-IR excess

3.3 Class I versus Class II YSOs

3.4 Mid-IR excess versus near-IR excess

3.5 The Class III YSO census in $\mbox{$\rho$ ~Ophiuchi}$