Based on observations made with the ESO VLT at the Paranal Observatory under programme IDs 63.I-0329 and 67.C-0264 and with ESO'S TIMMI2 on La Silla under programme ID 71.C-0438.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

All appendices are only available in electronic form at http://www.edpsciences.org

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... IDL

Interactive Data Language.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... S071-D

We mention here that, unlike the other finding charts shown in Persson et al. (1998), the finding chart for S071-D is (accidentally) not centred on the actual calibration star, this at first caused some confusion for our calibration.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Observatory

Observations at the Palomar Observatory were made as part of a continuing collaborative agreement between the California Institute of Technology, Cornell University, and the Jet Propulsion Laboratory.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... MIDAS

Munich Image Data Analysis System.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... conterpart

Persi et al. (2003) also identified the MIR source with the HMC on the basis of low signal-to-noise data and related it to the K band emission found by Testi et al. (1998). However, they apparently underestimated the significance of the multiplicity of objects in the G9.62+0.19-F region and thus could not distinguish between F1 (dominant in the K band) and F4 (dominant in the N band).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...2002a)

We mention our experience with another massive star-forming region (Apai et al. 2005) where PAH emission (also covered by the MSX A band) clearly affects the SED of the central object.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... astrometry

For the probable cause of the astrometric disagreement between our positions and the one of Persi et al. (2003) we refer to the end of Appendix A.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... vicinity

It is difficult to confirm this claim just by looking at Fig. 8. This kind of plot is well-suited for visualising the polarisation pattern for extended emission but is often misleading around compact sources with steep intensity gradients. In these cases the polarisation has to be derived by aperture photometry centred on the compact sources in all the polarimetric frames and then applying Eqs. (1) and (2).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... efficiency

Usually, the scattering efficiency is strongly diminished at wavelengths beyond $\approx$ $2 {-} 3 ~ \mu$ m when only dust is considered with the standard MRN size distribution (Mathis et al. 1977) typical for the interstellar medium, i.e., the overwhelming majority of the dust particles is sub-micrometer-sized with a typical diameter of $\approx$ $0.1 ~ \mu$ m.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... temperatures

Beckert et al. (2000) demonstrate that this approximation (Altenhoff + Mezger & Henderson correction factor) is in very good agreement with non-approximative numerical solutions up to frequencies of $\sim$ 100 GHz.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...2000)

The 3.6 cm map of Testi et al. (2000) has a higher SNR, but Hofner et al. (1996) found that the behaviour of the components D and E already deviates from the optically thin regime at this wavelength.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... line

The RRL velocity is confirmed by Lockman (1989) who gives $v_{\rm LSR}$ = ( $4.1 ~ \pm 0.8$ ) km s^-1 for the H85 $\alpha$ , H87 $\alpha$ and H88 $\alpha$ lines towards G9.62+0.19, based on observations with the NRAO 43 m telescope with a FWHP of 3 arcmin.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...1998)

We used the USNO-A2.0 catalogue because it is more or less directly tied to the HIPPARCOS reference frame. In the meantime, we also checked the astrometry for the G9.62+0.19 region according to the 2MASS All Sky Survey (Cutri et al. 2003). An offset of around 0.3 arcsec seems to exist between the two catalogues for G9.62+0.19. We can and will not evaluate here which catalogue is more accurate. We just mention that the 2MASS astrometry places our object F4 even closer (0 $.\!\!^{\prime\prime}$ 15) to the peak of the HMC molecular line emission (see Fig. 4).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.