PRESIDENTIAL ADDRESS. 



yourself to a feeling of antecedent probability, and your trust 

 will sometimes be betrayed." 



A Case of Discontinuity. 

 It is on record^ that in order to show how the continuity 

 of the longest sequences may be broken, Babbage devised 

 a machine, "luhich produced numbers according to a 

 particular law for an indefinite period, then broke this 

 uniformity by a single exception, and, thereafter reverted 

 for ever to its original principle of action.''' 



And so in all scientific work, although we may repeat 

 experiments again and again with identical results, this 

 may be for example, because some small quantity of 

 impurity was present (or absent) in all cases, or because 

 the surface of our vessels of experiment had in all cases 

 approximately the same amount and kind of other matter 

 absorbed at the outset; and possibly in the next time of 

 experiment, the conditions may have altered, and the 

 results with them. Any experimenter with a little experi- 

 ence is so well aware of these facts that he finds it advis- 

 able in some experiments to compare the results with 

 other "check" experiments, carried out, if possible, on the 

 same day. In this way we try to do all that is possible to 

 guard against any unthought of condition arising as the 

 experiments are in progress. It often strikes us as very 

 strange that we may repeat an experiment many times 

 with a certain definite result, and then on a further repeat 

 obtain a slightly different result which after some reflection 

 we may be able to ascribe to a slight alteration in pro- 

 cedure. It is in this way, however, that new discoveries 

 are made. 



Repetition of Experiments. 

 I believe the statement may well be made that the 

 repetition of an experiment is never superfluous. When 

 the personal factor comes into play as in the practice of 



