1921.] 



Seed Control, 1920-21. 



817 



in the 1918-19 season, the discrepancies were mainly in the 

 •lover and grass samples. 



The principal sources of error in the statements made by 

 yendors of seeds during 1920-21 are given below, together with 

 the corresponding figures for the previous season : — 



Correspond auj 

 1920-21. Ji<jures 



1919-20. 



(a) Number of cases in which the declared percentage 



as to germination differed from the figures 

 ascertained on the official test — 



between 10 per cent, and 15 per cent. — 27 ... 17 

 „ 15 „ 20 ,, —32... 5 



over 20 per cent. — 53 14 



(b) Number of cases in which the difference in purity 



was from — 



3 per cent, to 5 per cent. — 3 15 



5 per cent, to 10 per cent. — 2 ... ... 6 



Over 10 per cent. — 1 4 



(c) Number of cases in which the presence of Dodder 



was not declared — 20 20 



(d^ Number of cases in which the percentage of 



injurious weed seeds was not correctly 



declared — 7 32 



(e) Number of cases in which the declaration was 



incomplete — 7 5 



it) Number of cases in which no declaration at all was 



given — 25 7 



It will be observed that the number of cases in which the 

 percentage figure of germination was radically wrong has in- 

 creased by over 300 per cent, as compared with last year. Of 

 the total number of cases in which the declared percentage of 

 germination was at variance with the results ascertained on the 

 rests, namely, 112, 69 related to clovers and grasses, and 41 

 to vegetable seeds. Of the former, trefoil showed the worst 

 results, 18 cases being recorded of which 14 showed a difference 

 in germination of over 20 per cent. Among the vegetable seeds 

 parsnip was the worst offender, accounting for 12 cases of which 

 7 showed a difference of over 20 per cent. 



As experience in the administration of the Testing of Seeds 

 Order increased with each year's working, the Inspectors were 

 careful to draw control samples only from bulks which they had 

 reason to think were not up to the standard declared by the 

 vendor. It is, therefore, a difficult matter to draw general con- 



p 



