BY comparison with the voluminous literature on the theories of the firm and consumer, the amount of space devoted to the theory of the union is small indeed. This is not accidental; it results from the fact that the behavior of firms and consumers can be easily interpreted as maximizing while that of a union cannot. In recent years, however, the theory of the union has received increased attention. For example, John T. Dunlop in his now famous volume of essays, Wage Determiinationi Under Trade Unions,' has attempted to apply the maximization principle to labor union behavior, arguing that a union typically, although not universally, attempts to maximize the money income of its membership. The thesis that unions are maximizers has served as a spur to the enunciation of at least one antithesis: Ross's theory that union behavior is determined by considerations.' Ross's view of union behavior is not entirely new, but he has given the most lucid statement of this position of which I am aware. However, his arguments have not convinced Dunlop; in the preface to the new edition of Wage Determination he argues that the importance of the political determinants of wage rates has been exaggerated by Ross and others, and that political factors (although occasionally inmportant) are, on the whole, a relatively minor factor in wage determination.3 The issue is thus joined and a rather large-scale controversy is in the making;4 the present paper is offered as a contribution to this debate. Before embarking upon the discussion proper it nmight be well to define some of the crucial concepts used. When I speak of the political aspect of union behavior, or the political determinants of such behavior, I refer to those aspects of the behavior of individuals acting in behalf of a union ' which are best understood if it is assumed that they are motivated by a concern with the interests of the union qua organization,6 rather than with the interests of its members considered as individuals. It is not alleged that the interests of a unioil organization and those of its members are always or usually in conflict; indeed, the frequent confluence of these objectives has been a source of considerable theoretical difficulty. The individuals who act on behalf of a union -hereafter called union leaders -are alleged by Ross to be best understood as seeking primarily the well-being of the union organization rather than the welfare of its members. The well-being of the union organization, as I see it,7 involves the attitudes and behavior of three groups of people toward the union organization: (i) the members of the union,8 (2) the employers of the union members, and (3) the community at large, especially as its attitude is manifested in the behavior of the government toward the union. It is not seriously denied

OR many years there has been deep concern as to whether a shortage of doctors exists either nationally or in particular localities. This study investigates how well the distribution of the national stocks of medics (the generic term used here to refer to physicians and dentists) among areas corresponds to the distribution of population, and what influence is exerted by other variables such as effective demand for medical service, barriers to migration, and the locational preferences of medics. The unit of area in this study is the state; this is mainly because of availability of data. However, differences of size among the states, as well as in the relative populations of rural and urban areas and in the distribution of the urban population among large and small cities, seriously affect the observations. To remedy the resulting distortions as fully as possible, one of the authors is now analyzing the determinants of physician location among counties with results to be presented in a subsequent paper. However, many more variables for longer periods of time can be analyzed for states as a whole than for smaller sub-divisions.' The substance of our findings is contained in the regressions presented in the various tables. In considering them, the reader should bear in mind the following points: (1) We have analyzed a number of distinct types of medics: All Physicians, Self-employed Physicians and Dentists. Where our findings apply to all types we make generic statements. (2) To avoid discursiveness we discuss only those coefficients believed indicative of pervasive long-run forces affecting the allocation of medics. (3) In section I, the discussion is purely descriptive, i.e., the regression coefficients reflect observed associations but are not estimates of the structural parameters of a particular model. They do, however, have theoretical relevance. Some estimates of demand and supply functions are discussed in section II. (4) The years studied are 1930, 1940, 1950 and 1960 or, for some variables, years close to them. Lack of data for a sufficient number of years makes satisfactory time series analysis impossible. Section III summarizes our findings, and comments briefly upon their empirical implications. Our dictionary of variables is as follows: X (i)-Number of medics in the ith state. X2 -) Population of the ith state. X3()(Total) personal income of the ith state. X4-Volume of training facilities (number of places in medical classes) in ith state. X5-i Barriers to entry (percentage of applicants for licensure who fail examinations) in ith state. X6Population in ith state living in urban areas of more than 2,500 persons. X7-) Average income of medics in ith state.