It has been shown that many radio sources contain structure of size 1" to 10". The only practical way of making detailed brightness distributions of these sources is the construction or synthesis of a large phased array, with a resolving power of a few seconds of arc. It can be shown that the problem of phasing an array is essentially equivalent to the determination of positions of point radio sources, to an accuracy of about one- tenth of the array beamwidth. In an investigation of the phasing problems encountered with an interferometer, the positions of about 40 point radio sources have been determined, with an accuracy of about 1" in the best cases. The baseline constants of the interferometer, equivalent in concept and effect to the level and collimation errors of a meridian circle, were determined by requiring positions of several quasi-stellar radio sources (QSS) to agree in the mean with the measured positions of their optical counterparts. Although a fundamental coordinate system may be constructed from the radio observations alone, it is of great interest to compare the radio and optical systems in order to expose the systematic errors in both. It would perhaps be best to base the true positions on the optical system and to attribute any residual dif- ferences to instrumental errors in the radio system. In order to provide a well-determined fundamental coordinate system for the arrays which will be constructed in the next decade, optical positions of tens of QSS to an accuracy of 0'.'2 or better are needed. The use of QSS as a fundamental position system has the additional advantage, besides being available to both radio and optical instruments, of not suffering from the effects of proper motion.