122 THE AMERICAN JOURNAL OF SOCIOLOGY
such as parallel variations, is verj' largely a matter of accident, especially in the earlier stages of any causal analysis. In the most favorable cases, there- fore, much time must elapse in research. Moreover, there must be an exten- sive collection of material upon which observations can be made. (For instance, Darwin's labors for thirty years in collecting observations before he ventured his generalizations on the variations of species.)
It not seldom happens that approximation to the experimental method is estopped by the fact that the objects observable are too individual in their character, so that even the somewhat like phenomena are still too different for precise comparison. This is the case, for example, with those incidents of evolution in which periodical repetition is out of the question, at least in any period which we can observe. This is true of the origins of cosmic and organic structures, and of historical occurrences. In such instances com- parative causal analysis must be content in part with remote analogies, in part it must be guided by presumptions from a more general range of facts, from which application is possible to the object under investigation. Thus, for exam- ple, analysis of the origin of species falls back on observation of individual development and upon the demonstrable variations of artificially produced types.
In a different field, historical analysis follows generally recognized psy- chological principles in place of experimentation, and so reaches tentative conclusions.
The third stage is logical analysis. It consists in the separation of a com- plex fact into its component parts, with reference to the logical relations of the same. The presupposition of the process is that ideas have been formed of the qualities of the (distinguishable) elements which make up the whole under investigation. If this condition is fulfilled, logical analysis then under- takes to follow out the separate consequences which result from these qualities.
For example, legal conceptions may be cited. It is presumed that primary descriptive analysis has proceeded far enough to put the thinker into posses- sion of the separate notions involved in jural society. Logical analysis then undertakes to arrange these distinct notions in rational relations to each other, thus forming a philosophy of legal society.
II. SYNTHESIS.'
Synthethic procedure may consist in the simple reversal of a preceding analysis. In that case the synthesis is of a reproductive character. It has a relatively limited value, since it serves practically as a corroboration — or, as we say in arithmetic, a "proof " — of the analytic results. The synthetic process may, however, be so used that only certain results of the previous analytical investigations are employed, or, perhaps, only the elementary conceptions,
' Vide Small, "Sociology and Economics," Journal of Political Economy, March, 1895, pp. 179 seq.
