Some evolved stars exhibit peculiar features in their IRAS Low Resolution Spectra. The features are difficult to be assigned to any known atomic/molecular bands or a normal composite of grains. The peculiarities in the spectra can be classified into three types:(1) very sharp peaks at wavelengths which do not correspond with any known atomic or molecular bands, (2) a broad but apparently band-like feature at special wavelengths, (3) a very broad flat feature with a peculiar inclination in the whole spectra betweeen 7-25 micron. One possible interpretation of these strange spectra is that they are due to a malfunction (or noise) of the IRAS low resolution spectrometer at the time of data acquisition. The other interpretation is that the infrared source consists of two objects closely laying, and the spectra of the different types are symbiotic. The third possibility is that these are true anomaly of the stellar spectra due to unknown molecular or dust band emission/absorption, or atomic lines in unusual excitational states. First, we would like to confirm the peculiar features which were observed before with IRAS. Then, we wish to identify and search new features. They may be a true anomaly of the spectra, which may originate from unusually intense atomic or molecular lines (such as infrared lasers or Raman scattering), or from the emission band of unknown solid in the circumstellar envelope. One of the aims of this proposal is to find previously unknown, unusually strong lines of atoms or molecules in the mid-infrared wavelength range, and to use them as future probes of circumstellar envelopes.