be in
intimate association, is the same. The second conclusion is
necessarily dependent upon what we take to be "primary elements" and
"secondary elements;" and the question is how can these be determined?
As a rule it will be found that the primary elements are the most
constant parts of the whole group of examples, appearing more
frequently, possessing greater adherence to a common form, changing
(when they do change) with slighter variations; while the secondary
elements, on the other hand, assume many different varieties of form,
are by no means of constant occurrence, and do not even amongst
themselves tend to a common form. The primary elements, therefore,
constitute the form of the custom which represents the oldest part of
the survival. They alone will help us to determine the origin of the
custom, whether by features represented in the elements thus brought
together or by comparison with ancient custom elsewhere or with
survivals elsewhere similarly reconstituted. Altogether these
elements, thus linked together by the tie of common attributes, are
parts of one organic whole, and it is on this reconstructed organism
we have to rely for the evidence from tradition.

When any given custom or belief has undergone this double process of
analysis of its component parts and classification of its several
elements, another process has to be undertaken, namely, to ascertain
its association with other customs or beliefs, in the same country or
among the same people, each of which customs or beliefs, being treated
in exactly the same manner, is found to exhibit some degree of
relationship in origin, condition, or purpose to the whole group under
examination. In this way classification, analysis, and association go
hand in hand as the necessary methods of studying survivals. Without
analysis we cannot properly arrive at a classification; without
classification we cannot work out the association of survivals.

The process is perhaps highly technical and complicated. It may not be
of interest to all to discuss the process by which results are
attained when what is most desired are the results themselves. But in
truth the two parts of this study cannot well be separated. To judge
of the validity of the results one must know what the process has
been, and too often results are jumped at without warrant; items of
custom and usage or of belief and myth are docketed as belonging to a
given phase of culture, a given group of peop