The ratio Q of infrared to radio luminosity of galaxies dominated by star
 formation is a well-defined and universal quantity (Helou et al. 1985; de
 Jong et al. 1985).  Out of 550 galaxies brighter than 5 Jy at 60um, only
 two objects, NGC 1377 and IC 1953, are characterized by a ratio Q more than
 a factor of 10 and more than 6 sigma greater than the average value (Condon
 et al. 1990).  We propose to obtain photometry and spectroscopy with ISO on
 these two objects to help determine why they are different, and thereby
 gain more insight into the physics of the basic radio-IR correlation.

   Using ISO data in combination with ground-based data we hope to
 discriminate between two hypotheses that might explain this singularly high
 ratio: (1) an abnormally weak interstellar magnetic field, or (2) that we
 happen to observe the galaxies at a privileged, very early epoch in a
 star-formation burst.  Hypothesis (1) predicts an interstellar medium
 dominated by low-density ionized and photo-dissociated regions (PDR), with
 little mass in the cold, dense molecular phase; hypothesis (2) predicts the
 reverse.

   The atomic fine-structure lines accessible to LWS will yield physical
 parameters (density, pressure, T) for PDRs, while the ionic lines will
 probe the HII regions.  Long-wavelength PHT photometry will yield the dust
 temperature distribution.  These diagnostics will allow us to test the
 hypotheses above, both directly and by comparison with "normal" star
 forming galaxies observed under the Guaranteed Time programme.