The most striking results are i) the excellent correlation between CO and the dark lanes visible on the optical image (see figure on cover page), ii) the thinness of the molecular arms, iii) the good agreement between the CO and HI arm crests, iv) the detection of large streaming motions in the immediate vicinity of bright HII regions.
Figure: a- Relation between the CO and HI integrated intensities, I(CO) and I(HI), inside a 6 kpc-long, 1.5 kpc-wide lane centred on D 39 and following the southern R= 11 kpc arm of M 31. Each point corresponds to a arcsec cell (5 10 pc in the plane of M 31). b- Relation between the total hydrogen column density, estimated from I(CO) and I(HI), and the dust apparent opacity in the red.
It is often argued that the optically thick mm emission of CO does not trace the molecular gas properly and could be biased toward warm regions. The rich pattern of dust lanes visible throughout M 31's bright stellar disk yields another way to detect interstellar clouds, which does not depend on temperature or chemistry. Our high resolution CO map and the interferometric HI map of Brinks and Shane (A&AS 55, 1984) allow for the first time a detailed comparison of CO, HI and dust as tracers. All dark lanes with galactocentric radii kpc are detected in CO. CO is always brighter where the extinction is larger, while HI can be bright in regions with moderate visual extinction.
This is illustrated in Fig. 4a which compares point by point the CO and HI velocity-integrated intensities I(CO) and I(HI) in the conspicuous kpc arm. Whereas HI is bright in many CO-dim regions with moderate visual extinction, CO is never observed where HI is faint, or the extinction is low. We have plotted in Fig 4b the total hydrogen column density, derived from the expression , as a function of the apparent visual opacity, . N(H) is seen to scale with : the slope is H-atom cm , close to that for the clouds in the solar neighbourhood. The linear correlation coefficient is , significantly larger than that obtained by plotting I(HI) or I(CO) alone ( ). The good CO-dust correlation is a very strong indication that CO traces the H column density. It also implies that a sizable fraction of the gas in the arm is in molecular form.
Nature in press. Pre-/reprints: email@example.com and firstname.lastname@example.org.