88 STUDIES IN SEEDS AND FRUITS 



their original weight by talcing up water from the air. Of two seeds 

 broken up and exposed, the one to the ordinary air of a room, the 

 other at first to a temperature of loo^ C, with the resulting loss of 

 6 or 7 per cent, of its weight, both will ultimately be found after a 

 lapse of a few days to have acquired a weight on the average 1 1 or 1 2 

 per cent, in excess of their original weight in the entire condition. 

 On the other hand, the permeable seed of Canavalia ensiformis displays 

 quite a different behaviour after being exposed to the oven test. In a 

 week or two it returns to its original weight, having regained all its 

 lost water from the air (15 per cent.), and then maintains a stable 

 weight subject only to the ordinary hygroscopic variation of 2 or 3 

 per cent. (p. 78). 



(8) It is estabUshed from the foregoing results that the capacity of 

 an impermeable seed for absorbing water from the air in the broken 

 state varies inversely with the amount of the water-contents in the 

 entire condition (p. 80). 



(9) It is indicated by the preceding observations that the ultra- 

 dryness of the seeds of Guilandina bonducella and of the impermeable 

 seeds, of which it forms a type, is maintained by the impermeabiHty 

 of the seed-coats. Though, as compared with permeable seeds, both the 

 coats and the kernel are ultra-dry, the coats contain more water than 

 the kernel (p. 81). 



(10) It is suggested on a priori grounds that negative germinative 

 results, even those obtained by investigators with the most modern 

 means of research, are not to be relied upon as concerning the longevity 

 of impermeable seeds, provided that the seeds tested are normal in 

 appearance and in weight and display the normal passivity of seeds of 

 the type. It is considered that a much safer test is to be found in the 

 durability of external qualities, all seeds to be regarded as potentially 

 living or " germinable " that are sound, impermeable, non-hygroscopic, 

 and of unchanging weight. It is held that the final proof of a seed's 

 vitality lies almost outside the scope of the direct experimental method 

 (p. 82). 



(11) Yet occasionally, through some defect in its coverings, the 

 impermeable seed, that appeared at one time as if it could live for ever, 

 begins to fail. This is shown in the gradual increase of weight extend- 

 ing over months, as a result of which the impermeable seed assumes the 

 r6le of a permeable seed. If this occurred under ordinary soil-conditions 

 the seed would have to choose between germination and death. If it 

 took place in a seed exposed to the air in a room, the seed might retain 

 its germinative power for some years, as was indicated in my experi- 

 ments (p. 83). 



(12) A speculative turn is given to the discussion by the reference 

 made to the " cosmic suggestiveness " of the ultra-dry condition of the 



