236 STUDIES IN SEEDS AND FRUITS 



2 or 3 per cent., being lightest in the summer and heaviest 

 in the winter months. Variable seeds, where there is a 

 mixture of permeable with impermeable seeds, display about 

 half this variation, namely, i per cent. But, as has been 

 said, the disturbing influence of the hygroscopic reaction is 

 a great obstacle in detecting small differences in such ex- 

 periments. If there has been a change, it has certainly not 

 involved any increase in the average weight. On the contrary, 

 the indications, such as they are, point slightly in the direction 

 of a diminution ; but it will not be possible to obtain definite 

 results until the experiments have been greatly prolonged and 

 one is able to eliminate the effect of the hygroscopic reaction 

 by comparing the averages for groups of years. 



In such experiments on hygroscopic seeds it is necessary 

 that they should always be kept in the same room. So 

 sensitive were my seeds to change that a transference from 

 one room to another was sufficient to cause a variation of 

 i or even of i per cent. Several years ago MM. Van Tieghem 

 and Bonnier published some interesting results in a paper in 

 the Bulletin de la Societe Botanique de France (tome xxix. 1882), 

 some of which bear directly on the effect of time on the weight 

 of permeable seeds. They found that after two years in free 

 air, peas had gained about \\ per cent, in weight, haricots 

 about 2 per cent., seeds of a species of Vkia about i per cent., 

 and Ricinus seeds rather more, but the actual increase in the 

 last case is not given. From my own observations it would 

 appear that this rise in weight was within the ordinary 

 hygroscopic range, and that if weighed at another season all 

 the seeds might have displayed a decrease instead of an 

 increase in their weight. 



As indicated in Note 25 of the Appendix, the variation 

 in weight of completely air-dried seeds of Pisum sativum^ 

 Faha 'vulgaris^ and Phaseolus multiflorus during a period of 

 nearly fifteen months ranged from 2*6 to 3*6 per cent, of the 

 seed's weight. The variation in weight is also there given 

 for the seeds of Ricinus communis in the case of experiments 



