H. M. FucHS 225 



another part. The actual counting was done by taking a number of 

 eggs up in a pipette and spreading them in a line along a glass slide, 

 which was then passed under a low power of the mici'oscope. Every 

 •percentage 7-ecorded is calculated from a count q/'400 — 500 egys. Since 

 most of the fertilizations were made with less than 2000 eggs, the 

 percentages obtained by counting 400 — 500 of these should be fairly 

 accurate. Each percentage is reduced to the nearest whole number. 



The final result of these trials, in which each experiment was made 

 double or triple in order to test the degree of accuracy attainable, was 

 that the extent of the experimental error in the percentages might be 

 as much as 3 7, • It will be seen, however, that in the results of the 

 experiments, the differences between percentages to be compared with 

 one another were almost always considerably greater than the extent 

 of this error. There is one way, of course, in which an absolute control 

 can be kept of any experiment. Each experiment might be duplicated — 

 that is to say, each fertilization be made twice with the same quantities 

 of sperm and eggs — and the results rejected if they did not agree 

 exactly. This method was not adopted because time would not permit 

 of it in all cases, nevertheless a similar control was kept in many of 

 the experiments. Each of the fertilizations was made double, but 

 different quantities of sperm were added in the two cases. By both 

 giving similar differences in the percentages, the weaker-sperm and 

 stronger-sperm series in an experiment confirmed one another. This 

 method had the further advantage that there was much more prob- 

 ability of the experiment being a success. If in the weaker-sperm 

 series the sperm was so dilute that no eggs were fertilized at all, 

 probably the addition of a greater quantity to form the stronger-sperm 

 series would give percentages lying between and 100, and permit of 

 comparison. Similarly, if it happened that the stronger-sperm gave all 

 100 7,, fertilizations, the addition of less sperm-suspension to the water 

 with the egg.s to form the weaker-sperm series might give a result 

 lying between and 100 °/,^. If, however, both series lay between 

 and 100 °/^, they were a check on one another. 



The above outline gives the way in which the experiments were 

 carried out, and the precautions adopted to ensure the maximum of 

 accuracy in the results. There is another and more obvious source 

 of error which has to be guarded against in all experiments on fer- 

 tilization. Each experiment is made with the eggs from a single 

 individual and the spermatozoa from a single individual. The presence 



Jouiu. of Gen. iv 15 



