The goal of this proposal is to determine the chemical properties
 of the dust shells of obscured proto-planetary nebulae (PPN)
 through a study of their short-wavelength infrared spectra.  Spectral
 features in the 10 to 20 micron region indicate the chemical nature
 (oxygen or carbon-rich), and the strengths of the features relate
 to the physical properties of the shells.  A few carbon-rich PPN
 have been observed to show PAH features and an unidentified 21
 micron emission feature.  We propose to observe a sample of IRAS
 sources which have the expected properties of PPN and which have
 no visible counterparts (V>20 mag).  These we have located and
 observed from the ground, obtaining near- and mid-infrared
 photometry.  Essentially all previously studied PPN have visible
 counterparts.  Compared with these, we believe this sample of
 invisible PPN are obscured either because they had a higher mass-loss
 rate on the AGB or because they are less evolved post-AGB objects
 and their shells have not yet expanded sufficiently for visible light
 to escape.  In either case, this study will allow us to investigate
 a sample of PPN at a different state of evolution from other PPN
 studied, which have visible counterparts of G-A spectral types.
 This will advance our understanding of the range and evolution of the
 properties of circumstellar shells in this important evolutionary stage.
 Only ISO provides the spectral range and resolution necessary to carry
 out these observations.