FOOD, HIBERNATION AND MIGRATION IO5 



14th March, 1934, was a migration night, but the same date in 1932 

 was not. In 1934, there was much rain, but if rain was the cause of 

 the difference, it is not clear why the 21st March 1932 showed such a 

 great amount of migration, because, as on the 14th, there was only a 

 trace of rain. When we come to wind direction, we seem on firmer 

 ground. The 14th, 19th and 20th March 1932, all warm days, had 

 winds with htde or no south in them. On the 21st the wind turned 

 south, and there was a great migration, hi 1934 the pre-spawning 

 period was marked by three northerly winds and a calm. The 20th, 

 a night when there was very litde migration, was pouring with rain 

 and mild, but the wind was due north. In 193 1 there was a great 

 migration on the night of the 19th March. The weather was warm, 

 but dry. Wind was east-south-east. I do not suggest that there is no 

 connexion between wet and warm nights and migration, for there is a 

 loose correlation, but in the case of the frogs visiting Large Totteridge 

 it seemed to be necessary that the wind should be in the south for 

 migration over the road to be extensive, and this is just what the 

 smell hypothesis requires. 



In the following paragraphs, I give a more formal analysis. The argu- 

 ment is statistical in character, and it is impossible to relegate all of it to 

 Chapter 10, Appendix i g, where, however, more details will be found. 



In the paper, I pointed out that there was a significant tendency for 

 the frogs to migrate when the wind was blowing from a southerly 

 direction, across the pond to the road where the animals were killed, 

 but I qualified this by pointing out that this was not very good 

 evidence for migration up-wind. This was, indeed, the expected 

 direction for a migration directed by smell, but winds from tliis 

 direction are often wet and warm. I said that the unravelling of this 

 kind of data would need more information than was then available. 

 This is only partly true, and I now think that I was not sufficiently 

 enterprising in the analysis, and too cautious in the conclusions. 



In Fig. 26, wind frequency and direction is compared with migration 

 and wind direction, the same scale being used for each. It is 

 obvious that most of the migration occurred when the wind was from 

 the southerly directions. In Table 6 the same data are arranged in 

 Columns i and 2, and in Column 3 are shown the amounts of migra- 

 tion that would be expected if migration had been at random, and so 

 merely in proportion to the frequency of the winds at the period 

 studied. It is obvious that there was a large surplus of frogs migrating 



8-(T.9i4) 



