314 



Mr Fowler and Mr Lock, The origin of 



equal, is roughly proportional to the spin of the shell. We can even 

 go somewhat further on these figures, for we notice that the values 

 of Sq'/Q, show a well-marked constancy through the whole table. 



Table I. 



The next table, Table II, also contains values of Sq/Q. for each 

 group for which they have been deduced from the observations. 

 The mean values of Sq/Q, for all groups for each gun are the same, 

 confirming the more precise comparison of Table I. 



The main purpose of Table II is to exhibit the correlation (if 

 it exists) between the values of (f>Q for the various rounds of a 

 group. It contains the mean value of (/)q for each group, and the 

 "spread" of the values of </>(j in the group, the spread being the 

 smallest angle which will include all members of the group. 



The entries of the table show" unmistakable regularities. In 

 the first place the spread is in general far too small for one to sup- 

 pose that the observed values of (/)q represent a random distribu- 

 tion over the whole 360°. The groups of two rounds each provide 

 only shght evidence on this point. But even here there is some 

 evidence of correlation, for it is obviously equally probable that 

 the spread of a pair will or will not exceed 90° for random distribu- 

 tions of ^Q. Of the seven such groups, the spread of one only 

 (III 22, 23) exceeds 90° and the average spread of the seven is 

 only 50°. 



There are twenty-one groups in the table containing three or 

 more rounds. In a group of three rounds distributed at random, 



