l86 ECOLOGY AND LIFE HISTORY OF THE COMMON FROG 



variations in colonies of tadpoles, and shows that the different colonies 

 really did differ by far more than could reasonably be ascribed to 

 chance. Unfortunately the analysis of variance has a rather hmited 

 value in ecology. This is because it so frequently happens that the data 

 are not sufficiently symmetrical for the method to be used. Even if 

 one tries to collect data in a symmetrical form, the efforts are often 

 frustrated by gaps in the records. It may be very tiresome if a farmer 

 fills in a pond half way through our investigation, and so destroys 

 a set of observations carefully designed according to all the best books, 

 but we cannot stop liim. Nor can we order droughts and snowstorms 

 just when we need them. We must use methods that are not so 

 sensitive to irregularities, even if we lose information and have to 

 expend more labour. 



Correlation Analysis 



This is a most important ecological method, for it fulfils the need 

 just mentioned. Provided that numerical data exists in corresponding 

 pairs or larger groups, correlation analysis can be used even if the data is 

 fragmentary. Moreover, both graphical and mathematical methods 

 can be used. Chapter 8 is really one long example of what can be done 

 with this statistical tool. 



The Statistical Methods of Chapter 8 



Ezekiel's methods have, so far as I know, not been applied to an 

 ecological problem such as that of Chapter 8 and it may therefore be 

 of interest to give here a short account of the procedure adopted in 

 working up the Phenological Reports, especially as the full investi- 

 gation has not been published anywhere else. 



The Phenological Report was an annual publication of the Royal 

 Meteorological Society. In the early years of the frog-spawn records 

 these dates were not fully included in the annual report, and in the 

 1935 work I had to extract the details from the observer's original 

 returns. In later years, the Report itself contained the necessary 

 details. I was fortunate in having the help of Dr. Frazer to copy out 

 the data and supply the whole of it in an easily usable form. For each 

 spawn record seven meteorological readings were actually used, and 

 some more were collected but were not thought important enough to 

 use. These records came from the Monthly Weather Report of the 

 Meteorological Office. As attendance there to extract the data would 



