The global synthesis and global optimization methods developed recently are based on numerical methods of search of the global extremum of the merit function. Such methods are developed without paying due regard to image formation law and with no account of the role of the separate optical subsystem for appearance of aberrations. These circumstances don't make possible effective formalization of the problem of obtaining the starting system. It should be mentioned that 'before the computer epoch' the optical system design was characterized by wide use of analytic methods based upon classical aberration theory. These methods were used both for optical system designing and the detailed study of their potential possibilities. The efficiency of said methods is proved by the possibility for an optical designer to get almost 'by hand' a real optical system. The labor-consuming and tedious procedure of the real system aberration correction took place only at final stages of operational development of the optical systems. The paper describes the author's methodology of starting system synthesis. This methodology makes possible numeric- and analytical-construction of optical systems on the basis of wide use of the classical theory of aberrations and it substantially develops 'before the computer epoch' approaches. Let us consider a set of principal peculiarities of the offered methods of starting system synthesis and the results of particular use of such methods.