1 62 STUDIES IN SEEDS AND FRUITS 



one would look for if the experiment was continued for a 

 year. A chance observation led me to believe that the whole 

 range was quite twice as much as that indicated in the fore- 

 going experiments ; and I soon found, on transferring the 

 seeds from a cool, damp room without artificial heat to a dry- 

 cupboard, where the temperature (owing to the vicinity of 

 hot-water pipes) was from io° to 15° F. warmer, that the 

 range of the hygroscopicity was much increased. In such 

 circumstances the seeds of Anona^ Shaddock, Canavalia ensiformis, 

 etc., which, as first tested, exhibited a variation in weight of 

 I to i|- per cent., now showed an increased range of 2 or 3 

 per cent. Accordingly, I subjected the seeds to this new 

 proof, and the results are those given in the following pages. 

 The seeds were transferred from the cool room to the warm 

 cupboard for two days and back again to the room, the mean 

 of the two results being taken. Typical impermeable seeds 

 used as checks in many experiments displayed no change in 

 'weight, except of a trifling nature. 



Results of the Author's Observations on the 

 Hygroscopicity of Seeds. 



I. Impermeable Seeds. (Hygroscopicity "nil.") 



A typical impermeable seed exhibits no change in weight, 

 except such as can be attributed to instrumental error, or to 

 a little loose tissue adhering to the scar or to the raph6. For 

 example, a seed of Guilandina honducella 40 grains in weight, 

 and a seed of Entada scandens weighing 400 grains, would 

 show the same variation during a month of o"i grain, which 

 is so small that it may be safely attributed to other causes 

 than to normal hygroscopicity. 



The seeds actually tested by me include the following : — 

 Ademnthera pavonina, Dioclea reflexa, Entada scandens, Guilandina 

 honducella, Leucana glauca, Mucuna urens, etc. But of course 

 all the seeds mentioned in the Impermeable Group in 

 Chapter V would, when typical, display no hygroscopic re- 



