In the first few years following the discovery of discrete cosmic X-ray sources, about three dozen objects were detected (Friedman, Byram & Chubb 1967; Fisher, Johnson, Jordan, Meyerott & Acton 1966; Clark
et al
. 1965; Gursky, Gorenstein & Giacconi 1967; Bradt, Naranan, Rappaport & Spada 1968; Cooke, Pounds, Stewardson & Adams 1967; Giacconi, Gorenstein, Gursky & Waters 1967; Chodil
et al
. 1967
b
). They represent a brightness range of about a thousandfold from the most intense source, Sco XR–1,
ca
. 5 x 10
-10
J m
-2
s
-1
, to the weakest sources at a few times 10
-13
J m
-2
s
-1
. The lower limit of detectability is inherent in the short duration of rocket flights and the limiting size of instrumentation that can be carried in vehicles like Aerobee, Skylark and Nike Apache. As a result of this instrumentation barrier, the pace of discovery has now slowed down considerably, and the next great surge in detection of new sources will have to await the use of satellites. Spectral information is still based on the relatively crude resolution of proportional counters and scintillation counters, so that no conclusive determination of thermal or non-thermal mechanisms is yet possible. Because position data are obtained with the use of mechanical collimators, accuracies of a few minutes of arc are the best that have been achieved, and only for a few sources. As a consequence, only a few X-ray objects are reliably identified with optical and radio counterparts.