296 THE INDIAN JOURNAL OF VETERINARY SCIENCE AND ANIMAL HUSBANDRY [ IV, IV.

gators like Kellner and Armsby have described the estimation of clover hay and
meadow hay respectively on this basis and their statements do not refer to any
serious difficulty, although with a combined feed the latter has remarked that " the
determination of digestibility of a concentrate in this way is less accurate
and the range of uncertainty thus introduced may be very wide". The present
authors have found that apart from such difficulties (in combined feeds) the results
obtained with a single feed like paddy straw also have varied widely; and in the
case of the protein fraction the matter has been further complicated with a large
number of negative values. With the addition of a concentrate and working with
assumed values of published standards for the latter, the results have also not been
satisfactory; and as will be explained later on, it seems doubtful if outside standards
are suitable under the conditions of our experiments.

These difficulties led to the trial by another method of experimentation based on
the graphical analysis of the data; and the results so far appear to hold out
a promise of a greater degree of accuracy. As a natural corollary to the graphical
solution, multiple regression equations were later on used at the suggestion of
Professor P. C. Mahalanobis and Mr. S. Bose of the Presidency College, Calcutta,
and the present paper deals with the various aspects of these results.

It is necessary at this stage to touch briefly on some of the salient points
leading to the initiation of the methods.

With the start of the Animal Nutrition Section at Dacca by a grant from the
Imperial Council of Agricultural Research, India, the first attention was necessarily
directed towards the behaviour of paddy straw since it formed the staple fodder of
the province.

Experiments were conducted on bullocks which were more or less representative
of the type of animals used throughout the province. The reason for this was that
any results obtained which could be reasonably accepted as reliable could then be
applied to the cattle population of the whole province.

As pedigreed animals are rare in Bengal (even the Departmental herd might not
yet be accepted as such) and also as such animals are somewhat inclined to delicacy
in feeding it was considered that it was better to use the ordinary country animals
than to have recourse to pedigreed herd of the Department.

Experiments with paddy straw as the sole feed as well as in combination with
other feeds were started. The object was amongst others to determine the diges-
tibilities of the different feeds, both singly and in combination by the method in
vogue, viz., the estimation of the digestibility of the single feed, say, paddy straw
at first, then repeating the experiment with the addition of a concentrate, say,
linseed cake, the computation of the second being made by a corresponding deduc-