292 



SCIENCE 



[Vol. V., No. 114. 



as to which the Indians declare that the island 

 was once the cemetery for the neighboring 

 mainland, and therefore will not allow the 

 mounds to be explored. 



Subsequently, near the village of San Miguel, 

 Mr. Miller examined and photographed the 

 ruins of an old church, sur- 

 rounded b} T a pavement of 

 smooth, flat stones, carefully 

 laid in cement, but now cov- 



*^i> 



lll'INS OF A TKMI-LE ON IUK 



ered with earth. The inhabitants say that this 

 pavement extends for half a mile around the 

 church, and that a broad, paved way once led 

 from the church to the water, a mile away. 



ERRORS IN DIGESTION EXPERIMENTS. 



Henneberg and Stohmann, in their Bei- 

 trage zur rationellen fattening der wiederkciuer , 

 published in 1800, reported practically the first 

 determinations of the digestibility of the proxi- 

 mate constituents of cattle-foods. Since that 

 time, a large number of similar determinations 

 upon various fodders, and with the several 

 species of domestic animals, have been made, 

 chiefly, if not entirely, b}- the German experi- 

 ment-stations. In these determinations the 

 method employed by Henneberg and Stoh- 

 mann, and which is here given in outline, has 

 been universally followed. 



The food of the animal is weighed, suitable 

 account being taken of any portion left un- 

 eaten, and a sample of the food is subjected 

 to chemical analysis. The solid excrement of 

 the animal, which consists for the most part 

 of the undigested portions of the food, is also 

 carefully collected, weighed, and analyzed. 



From these data, it is a simple matter to com- 

 pute how much dry matter or how much of 

 any particular ingredient of the food the ani- 

 mal received, and what part of this failed to 

 be digested. 



This method of experiment evidently will 

 give directly the digesti- 

 bilit} 7 of any fodder which 

 can be made the exclusive 

 food of the animal. In 

 the case of material like 

 grain, meal, and the con- 

 centrated fodders in gen- 

 eral, the matter is not quite 

 so simple. In this case it 

 is first necessary to deter- 

 mine the digestibility of a 

 sample of hay, or other 

 coarse fodder. This 

 done, the animal is 

 given a mixture of this 

 coarse fodder and the 

 concentrated fodder in 

 question, and the 

 amount of this mixture 

 which is digested is de- 

 termined. Then, on 

 the assumption that the 

 same proportion of 

 ^ the coarse fodder 

 ---^^ was digested in 



the second trial as 

 in the first, we calculate how much of the con- 

 centrated fodder must have been digested in 

 order to yield the results observed upon the 

 mixture. 



Certain sources of error have been ignored 

 in the general statement given above. Thus 

 the excretion is always more or less irregular 

 from day to day ; and the excreted matter con- 

 tains, in addition to undigested food, more or 

 less intestinal mucus, and remnants of digestive 

 juices, which, though small in amount rela- 

 tively, are not entirely to be neglected. Then 

 it has recently been shown that some portions 

 of the food fail to appear in the excreta, 

 because they suffer a fermentation in the ali- 

 mentary canal, rather than because they are 

 digested in any proper sense. This is particu- 

 larly the case with cellulose (see Science, No. 

 100, p. 11). Finally, the methods of analysis 

 in use for fodder and excrement are not in 

 all respects capable of giving sharply defined 

 results. 



Another class of errors, the small unavoid- 

 able errors of weighing and chemical analysis, 

 usually less considered, may grow to very con- 

 siderable dimensions when multiplied many 



