There has been a growing realization that ellipticals and lenticulars have a significant interstellar medium, and that interesting amounts of star formation take place there. There has been a decade of work on HI emission from ellipticals and more recently CO observations in several cases (see e.g. the reviews by Wardle & Knapp (1986) and Schweizer (1987)). Although the majority of the galaxies detected by IRAS are spirals, quite a number of ellipticals and lenticulars were detected (e.g. Jura 1986, Jura et al 1987, Knapp et al 1989). Thronson and Bally (1987) have studied the IRAS colour-colour diagrams for these galaxies and conclude that they occupy the same region of the diagrams as spiral galaxies (and, for that matter, star forming regions in our Galaxy). About 2/3rds of the sample they studied have the colors characteristic of cirrus and 1/3rd those of dusty regions surrounding young stars. The star formation rate they derive (0.1-1 M0/year) is comparable to the mass-loss rate for evolved stars in these galaxies, but mergers and gas infall may also contribute significantly. Bally and Thronson (1989) studied the IRAS data for a sample of 74 S0 galaxies which had known single-dish radio fluxes. 30% were detected in all 4 IRAS bands and 80% were detected in at least one band. The galaxies divided into those which followed the infrared-radio relation for spirals, for which the radio emission is presumably due to normal star formation, and those with excess radio emission, presumably due to an active nucleus and jets or lobes. A small number showed a slight excess of infrared to radio. Similar conclusions were reached by Walsh et al (1989). Knapp et al (1989) report that 2/3rds of a sample of several hundred S0s are detected by IRAS at 60 and 100 µm.
Thronson et al (1989b) examined the IRAS data for 150 lenticular and elliptical 'shell' galaxies (Malin and Carter 1983), which are believed to be the result of low velocity mergers. Although some of the galaxies showed evidence for enhanced star formation, the majority did not and they concluded that either (1) the merging galaxies are almost always E or S0 with only modest amounts of interstellar gas, or (2) the time-scale for creation and maintenance of the shell is longer than the timescale for the starburst event, or (3) the formation of a shell structure requires a mass difference between the galaxies of a factor 10-100, so only a small fraction of the i.s.m. is heated or participates in star formation.
Walsh and Knapp (1990) find that the ellipticals detected by IRAS tend preferentially to be those with dust lanes visible in the optical. However the infrared properties are not strongly dependent on the visible dust content. They also find a slightly enhanced 100 µm detection rate for ellipticals with shells, boxy isophotes or inner discs, all of which are evidence of a recent merger, a result which is not necessarily inconsistent with that of Thronson et al (1990).
It is unfortunate that the infrared sources associated with early type galaxies are almost all rather weak, so that there is little immediate prospect of detection at wavelengths > 300 µm, and hence of accurate dust mass determinations.