Astron. Astrophys. 330, 443-446 (1998)

3. Results

In Fig. 1 we present our JCMT spectrum, together with the CO data from Frayer et al. (1994). The overall rms is mK, or 12 mK if we limit ourselves to the central portion of the spectrum. The resulting upper limit on the integrated line intensity, calculated using the formulae derived by Seaquist, Ivison & Hall (1995), assuming a rectangular profile with FWHM 680 km s^-1 (similar that of the CO(1-0) line reported by Frayer et al. 1994), is K km s^-1. Frayer et al. (1994) give (CO(1-0)) = K km s^-1, or 10.0 Jy km s^-1, or W m^-2, hence the measured [C II ]/CO(1-0) intensity ratio is .

Fig. 1. Spectra of the DLAAS towards PC1643 4631A (this paper; Frayer et al. 1994). The zero point of the velocity scale represents the expected position of the plotted lines for . Top: NRAO 140-ft CO(1-0); middle: NRAO 12-m CO(3-2); bottom: JCMT 15-m [C II ], with km s-1 corresponding to 459.399807 GHz. The spectra have been binned to 40 km s-1 in all cases. Zero-order baseline corrections have been applied.

Braine et al. (1996) reported an rms of 1.25 mJy in channels of width 224 km s^-1 for their observations of CO(3-2). This translates into a limit on the integrated line intensity of Jy km s^-1 or W m^-2 (assuming a rectangular profile with FWHM 800 km s^-1, as reported by Frayer et al. 1994) - a factor of 7.1 lower than the integrated CO(3-2) line intensity ( Jy km s^-1) reported by Frayer et al. (1994). Assuming that the CO(1-0) has been similarly overestimated, this yields a [C II ]/CO(1-0) intensity ratio of .

Fig. 2 shows a graphic demonstration of the dangers of coadding overlapping spectra to improve velocity coverage. In this case, baselines were not subtracted from the individual segments. The effect of combining the poor baselines is to generate a very convincing emission feature (still more so if we subtract a linear baseline at this stage - see the lower panel of Fig. 2). The apparent emission line is centred at km s^-1 for and has a full width similar to that of the controversial CO lines, though more Gaussian in profile.

Fig. 2. Top: Spectrum (dashed line) of the DLAAS towards PC1643 4631A, with km s-1 corresponding to 459.399807 GHz ([C II ] for ), constructed from individual 600-km s-1 segments (solid lines) without correcting their baselines (following the method of Frayer et al. 1994). Bottom: The combined spectrum after subtracting a linear baseline.

The integrated intensity of the apparent line is K km s^-1 on the scale, which would have indicated a [C II ]/CO(1-0) intensity ratio of 63000. This would have led us to completely the wrong conclusion since this value is consistent with those of low-metallicity systems (see the discussion that follows). Moreover, this would have been regarded as strong support for the validity of the CO detections and as indicative of rapid ongoing star formation in the DLAAS towards PC 1643+4631A.

Offsets such as those seen in the upper panel of Fig. 2 are usually the result of incomplete sky subtraction, or poor instrumental stability. We suspect the former in this case, even though the spectra were obtained during excellent and seemingly quite stable conditions. It is possible that more frequent nodding between the signal and reference beams would have reduced the offsets, but such anomalies are a fact of life in the submillimetre regime and we can be grateful to some extent that the baselines produced RxC2 and DAS are such good approximations of zeroth order. There are no fool-proof methods of achieving perfect sky subtraction and if there is a lesson to be learned, it is that high-bandwidth receivers and spectrometers are extremely desirable in this field.

© European Southern Observatory (ESO) 1998

Online publication: January 16, 1998
helpdesk.link@springer.de