THE FOODS FED TO BEASTS 189 



Such a calculation assumes that the figures apply to the 

 particular case in question. The full table given in Kellner's 

 work supplies a considerable amount of information which 

 permits one to apply these values with a fair degree of 

 certainty. There is, however, always the difference between 

 the actual conditions prevailing and those under which the 

 tables were deduced. A study of the tables in Kellner's 

 work shows that in some cases very wide variations in the 

 results were obtained. The variations compensate for one 

 another to some slight extent . Probably the great variations 

 that may be observed in the digestible fibre are really 

 attributable to the fact that some of the materials which 

 are possibly called " fibre " in the solid excreta of the beasts 

 are really bacterial residues. The fluctuations observable 

 in the column " total matter digested " are more valuable 

 in assessing the probable error in these experiments . A study 

 of the tables will convince one that the use of these tables 

 will give a figure for the digestible ingredient per cent, which 

 is true to two or three units, but cannot be considered as 

 being any closer than that. In some instances it is quite 

 obvious that Kellner himself recognized that the figures of a 

 few experiments are not very reliable. It will be noted, on 

 referring to p. 388, that Kellner gives digestible coefficients 

 for "palm nut cake" and "palm nut meal, extracted/' 

 which differ from one another to a degree which is difficult 

 to credit ; but when he makes use of these figures for compiling 

 the table on p. 377, he uses for calculating the digestible 

 nutrients in those two substances, not the figures he has 

 himself quoted, but the average of the two cases. That is to 

 say, in calculating the digestibility of palm nut cake, he 

 does not use his own figures but an average obtained from 

 palm nut cake and palm nut meal. This procedure is quite 

 legitimate, of course, but shows that Kellner did not himself 

 attribute to his own work that degree of precision which is 

 sometimes assumed by those who use his tables. Such 

 apparent discrepancies in the table of digestibility of decorti- 

 cated cotton seed meal, where the digestibility coefficient 

 of the fibre varies from o to 100, though appearing very big, 



