22 



during the sifting, the several parts must he covered with watch-glasses. The loss of 

 weight is checked. To this end it is necessary to determine more or less accurately 

 the percentage of water of the sample, ohserving however the rule laid dpwn here 

 helow suh 3. This will facilitate the determination of losses of weight. If they exceed 

 certain limits, the test must he repeated. As limits we have adopted: 4% for, seeds of 

 which 2 grammes or le§s are sifted for determination and 2 "/o, if the quantity exceeds 

 2 grammes. Two parallel-tests are estimated to cover each another, if the results do 

 not deviate more than 2 % in the case of fine grass-seeds and more than 1 % in all 

 other cases. 



3. Previous drying of seeds with a rather high percentage of water. 



In order to keep the percentage of water within reasonable bounds, it is necessary 

 to previously dry the samples in case the seeds. prove to he insufficiently dry. When 

 the quantities are small and not too moist, it will- he sufficient to expose the seed to 

 the air for one or two hours at room-temperature and spread out thinly. If, however, 

 the seeds are very moist, they will have to be dried at a temperature, not exceeding 

 30° in a current of air and — eventually — afterwards exposed to the air to 

 become airdry. 



4. Appliance of translucent light. 



> All „seeds" which admit of recognizing the empty seeds by means of translucent 

 light, are examined in this way on the diaphanoscope-tahles, in use at my institute. 

 It must be observed that it will not do to assume for convenience sake certain pro- 

 portionate figures for the percentage of empty seeds in the coherent parts of grass-ears. 

 In many cases they will have to be plucked out and examined separately. In order 

 to secure accurate results, complicated procedures ar§ to be followed, which may be 

 considered by some pai'ties to be rather doctrinaire. An example of them is to be 

 found in Appendix E. 



Sieves must not be used in examinations by means ef translucent light, unless we 

 are sure that the percentage of empty seeds or impurity will not undergo an alteration. 



5. Examination of seeds demanding a modification of the usual method. 



Poa trivialis e. g. may contain Poa compressa, Brassica Napus may be mixed with 

 Brassica oleracea, &c. In these cases the species being not easily separable are 

 provisionally reckoned to the pure seed. So the ''rough" purity of a mixture of 80 "h 

 Poa trivialis and 10 % Poa compressa and 10 % other impurities, will he 90 %. ^ Of 

 this imperfectly homogeneous "rough pure seed", 100 seeds are told off and determined 

 carefully one by one. In the report of the given sample, it is stated that the purity 

 amounts to about 90%, but that there are 10 seeds of Poa compressa in every 100 

 pure seeds. In calculating the "approximate use-value" in this case, we proceed 

 from an "approximate" purity-figure of 90%. This way of proceeding does not guarantee 

 an accurate percentage, but it will prevent differences in the results, which would 

 otherwise be altogether unavoidable. Only in this manner f. i. is it possible to test 

 Brassica-mixtures or some Beta-mixtures that might want a complete cultivation on 

 the field. 



An analogous method is to be followed for the valuation of the percentage of 

 diseased seeds, which is made at our institute chiefly for corn and leguminous plants. 

 We have now considerable experience in this matter. In many cases we are able to 

 distinguish at sight diseased seeds from sound ones, and we have used our best 

 endeavours to obtain sound "pure seed" for the determination of the germinating power. 

 Much can be done in this respect with a highly trained staff and a good deal of practice. 

 Just these conditions however are apt to lead to dissimilar results, which we should 

 wish now to eliminate if possible. Finally I have come to adopt a very radical precept. 

 In determining the purity of corn or leguminous plants, no attention is paid to the 

 state of health or to the presumable state of health of the seeds, eyen though they 

 should be recognizable by a practised expert. All the seeds that — apart from the 



