[ 2 9 ] 

 DIAGRAM. 





If the whole ico plants were taken up and weighed, it is 

 probable the experiment would be the more accurate; but 

 fhould that trouble be thought too great, the fixty plants 

 contained within the inner Tine of fmall dots would anfwer 

 perfectly well; or mould that bethought too many, flill 

 the 16 plants in the diagonals, with four near the centre 

 marked alfo with dots, fo as to make in all twenty, would 

 anfwer the purpofe fufficiently. All that is here required 

 is, that a certain order of felecHon fhould be previoufly 

 adopted, and mod ftridtly adhered to ; for mould a random 

 fele&ion of a certain proportion of the plants be permitted, 

 this might be done in a particular manner either through 

 prejudice or favour, which might affect the accuracy of the 

 experiment. — A fquare fpot divided in this manner fhould 

 be fet apart for each of the undermentioned diftances be- 

 tween 



