TRANSACTIONS OF SECTION L. 749 
11. Analysis and Synthesis in Learning Processes: 
By Dr. E. O. Lewis. 
This paper gave an account of a few initial experiments of a somewhat 
extensive investigation into the respective réles of analysis and synthesis in 
learning processes. Most of these experiments were performed with subjects 
aged eight to fourteen. In almost all cases the subjects were tested individually, 
but a few ‘mass’ experiments were performed to verify some of the chief results. 
The first series was designed for the purpose of finding the relative ease 
and efficiency with which problems are solved or material learnt when subjects 
proceed (1) from the whole to parts (W method), or (2) from the parts to the 
whole (P method). 
(a) Subjects learned to spell strange words which were exposed letter by 
letter (P method), and another series of words which were apprehended as 
wholes (W method), the whole word being exposed for the same period as it 
took to expose a corresponding word letter by letter—usually from ten to fifteen 
seconds. The data obtained showed conclusively the superiority of the W 
method, but this superiority was more marked in the complete learning of the 
matter than in obtaining first impressions. The P method compared more 
favourably in the learning of Welsh words, e.g. ‘ynghwydd,’ than in learning 
words constituted of familiar English syllables, e.g. ‘stimrectwomrud.’ 
(6) In this set of experiments words were exposed in syllables, each con- 
sisting of three to six letters (P method), and the results compared with those 
obtained by learning similar words by the W method. The syllabic presentation 
was the better both as regards the first impressions and complete learning; but 
this superiority of the P method was not so marked or consistent as that of the 
W method in series (a) experiments. 
(c) Drawings of familiar (e.g. ship) and unfamiliar (e.g. irregular designs) 
objects were observed and the subjects were asked to draw from memory what 
they could remember. One set of drawings was observed as wholes, whilst 
another set was observed in parts. The drawings in the part series were 
constructed by drawing the object in broad outline first and filling in the details 
in the succeeding drawings. The reproductions were assessed from the point 
of view of accuracy and completeness of observation. The results proved the 
superiority of the W method in a striking manner, especially the reproductions 
of the familiar objects. 
(zd) Common regular geometrical figures (e.g. oblongs, circles) were cut in 
cardboard into four, five, or six irregular parts. The subjects, mostly adults, 
were asked to reconstruct the whole by setting the parts together (P method) ; 
and to do the converse with another set, by indicating on paper how a given 
regular figure could be divided into a number of parts which were shown to 
the subject (W method). The times required to solve the problems were 
obtained by means of a stop-watch. In all cases the W method proved much 
the easier procedure. 
The contrast between the results of series (6) and (c) suggested that the 
superiority of the analytic procedure lies chiefly in mastering organic wholes, 
whereas the synthetic procedure would prove the better with mechanical wholes. 
A poem would form an organic whole as compared with a list of nonsense 
syllables which really only form a mechanical whole. This suggested the 
following experiment :— 
(e) A list of associated words, e.g. sun, warm, cold, was learnt by the entire 
method, and a similar list by the sectional method; and likewise with two lists 
of nonsense syllables. The entire method proved the more economical with both 
sets of material, but more so with the associated words. The results suggest 
that to attribute the superiority of the entire method to the saving of unnecessary 
associations, as is usually done, cannot be regarded as a satisfactory explanation, 
if indeed it does not overlook the most important factor. 
A result of some interest was obtained incidentally from one of the ‘mass’ 
experiments. A class of pupils was set the task of making as many sense words 
as possible in five minutes’ time out of the letters of some common word, e.g. 
treasure. The results obtained were correlated with the order of intelligence 
of the pupils as placed by the class-teacher. The coefficient of correlation was 
