The publication of the first and preliminary second catalogue of the COS-B sources, has brought to the attention of the astronomical community the new reality of the high-energy Galactic Gamma-Ray Sources (GGRS). It is worth recalling here the definition of such objects, in the COS-B sense: “A GRS is a significant excess of photon counts, compatible with the instrument's angular resolution, or, more clearly, with the instrument's Point Spread Function”. This definition of an unresolved (as it is the case for the vast majority of the GGRS) GRS is, to some extent, dependent on the shape of the source spectrum; however a general shape of the PSF for the case of the source associated with 3C273 can be found in Bignami et al, 1980. Naturally, the absolute flux (i.e. the total number of photons) from a source is also important in determining the positional error, especially when usage is made of the cross-correlation method. Typical photon numbers (100 Mev) for GGRS range from 50 to few hundreds. The COS-B catalogue error boxes are defined taking into account both the PSF and the photon statistics; it is then apparent that the search for candidate counterparts of the GGRS should be carried out inside such boxes, or in their immediate vicinity, and that, for examples, such loose positional coincidences as obtainable by increasing a quoted error radius by 50%, let alone doubling it, are totally unacceptable. Systematic searches for counterparts at other wavelengths have not yet been carried out, except for a few cases of X-ray and radio observations, as for example in the case of CG 135+1 (the only COS-B GRS erbox compatible with a 4U erbox) which led to the discovery of a new X-ray QSO by Apparao et al, 1978.