2 N. H. Exi'ERiMKNT Station 



.'111(1 87 \)or cent frerniination, the actual seeding value of 

 the sample is .9115 x .87, or 79 per cent. 



A certain amount of variation is to be expected in 

 comparing the results of two or more purity analyses. 

 I^'or this reason a "tolerance" or difference is allowed 

 between the averaj^e results of our tests and the j^uar- 

 antee in accordance with the followinfr formula. The 

 sample is considered as liciii^' made up of two jiarts, the 

 one beiiif!: the ])iu"e seed and the other the balance of the 

 sample. The tolerance in per cent allowed for the larger 

 jiart or ])ure seed is two-tenths of one per cent plus 20 

 per cent of the lesser i)art. For instance, if a sample is 

 guaranteed f)9.60 per cent pure seed, it nuiy test as low 

 as 99.32 i)er cent or as high as 99.88 per cent and would 

 be "satisfactory." In the same way, a sample guaranteed 

 92.80 i)er cent may test as low as 91.17 per cent or as high 

 as 94.43 per cent and be "satisfactory." If a sample is 

 found to be better than the highest figure in the tolerance, 

 it is marked "above" in this bulletin; if lower than the 

 lowest figure, the actual figure is reported. 



A wider tolerance for germination than foi- purity is 

 allowed between the results of the test and the guarantee, 

 as follows : 



Guarantee Allowable variation 



per cent per cent 



90 or over G 



80 or over but less than f)() 7 



70 or ovei- but less than 80 8 



(iO or over but less than 70 9 



Less than 60 10 



Accordingly, a sample guaranteed III* per cent geniiina- 

 ti(Mi is "satisfactory" it it tests S() pei- cent to 98 pei- cent. 

 A sample gnaran1ee<l S,") |)er cent may tall as low as 78 

 pel- cent or as high as 92 per cent. it' a s;iiiiple tests 

 higher than t lie i>er cent of variat ion allowed, it is marked 

 "above" in this bulletin, or, it' it tests below the variation 

 jdlowed. the actual figure is reported. 



