'I" 1 1 1-. A M I-; R I CA X l'.( )TA N I ST 59 



raceme, the Howers ot unv heiii.i;" a deej) retl while those of 

 the other were e(|ually hrilhaiU, hut vellow. 



In tlie autumn of 1917 another canna seed was sifted 

 from the sand pile alon^ with a much reduced showin<j^ of 

 seed remains; in fact the onl\- other distin<:]^uishable seed 

 coats were a few cucurbits, spinach and one each of Marty- 

 nia and winter vetch. The canna seed, having a rather 

 rough and dingy appearance was promptly germinated in a 

 flower pot. With the approach of frost the young plant was 

 taken indoors. The following winter being a severe one 

 the entire plant was killed above the surface of the soiU 

 though uninjured below. Early the next spring a new 

 growth appeared and after all danger of frost had passed 

 the plant was set out in the open. By early autumn it had 

 made a, rank growth and bore large deep yellow blossoms. 



The time required to obliterate traces of the seed coats 

 was undoubtedly extended by their being buried in a pile 

 above the surrounding surface from which all surplus water 

 could easily drain, yet the mound retained sufficient moisture 

 to support a heavy growth of grasses. In winter it was re- 

 peatedly frozen as hard as a rock during zero weather for 

 perhaps a foot from the surface. It is rather remarkable 

 that the canna seeds would have excluded moisture for so 

 long a time, thus preserving the vital parts against injury. 



Objection perhaps to the supposition that the canna seeds 

 found, iiad remained buried for so long a time may be made 

 on the ground that seeds from recently discarded tests could 

 have been carried and deposited in the sand pile by some 

 burrowing rodent. There is not much on which to support 

 such an objection since the sand tests made in 1901 were 

 discontinued at llial time .ind were followed in the interven- 

 ing years by an entirely different and cheaper method of 



