PERMEABLE AND IMPERMEABLE SEEDS 83 



but in my own experiments, where scraping or filing was 

 generally adopted, I attached but little importance to the 

 occasional negative results. Every seed of average weight 

 that possessed completely sound coverings, that displayed no 

 increase in weight whilst immersed in water for weeks, and 

 that behaved like a quartz-pebble in response to the hygro- 

 metric changes of the air, was for me a germinable seed, 

 whatever its age. An ineffectual result I regarded as merely 

 reflecting on my method, provided that the seed originally 

 possessed the qualities named. 



The final proof seems to me to lie almost outside the 

 reach of the direct experimental method. Nature would 

 probably follow a very slow and graduated process in pro- Suggested 

 curing the germination of ancient seeds ; and it is not easy longevity in 

 to see how we are to imitate such a process, and yet be theduration 

 confident that our failures lie in the seeds and not in the ternal 

 conditions of our experiment. In default of this there is impermeable 

 still open to us the plan of testing the duration of the three seeds - 

 chief qualities of an impermeable seed : the soundness of the 

 coats, the impervious character of its outer covering, and its 

 practically non-hygroscopic behaviour. The indirect method 

 would appear to give most promise in the investigation of 

 the longevity of impermeable seeds. 



Yet occasionally some strain of weakness due to a defect 

 in the shrinking process offers to the opposing external condi- 

 tions their opportunity, and the seed that appeared able to 

 survive for very long periods begins to fail. The indications 

 of the change are exceedingly interesting. They have already 

 been noticed on p. 76 in this chapter in the description 

 of the results of filing into seeds of Guilandina bonducella^ and 

 they will be again noticed in the account of a similar ex- 

 periment on the seeds of Entada scandens in Chapter VI. 

 During several months the filed seeds gradually increased in 

 weight, until, in the case of the first-named, they had increased 

 by 10 or n per cent., and in that of Entada scandens 2 or 3 

 per cent., assuming ultimately the condition of permeable 



