52 



On MendeTs Categories 



The probable error of all these values is of course consiilerable, but they tit the 

 observations as well as could perhaps be expected. Thus the lower limit of hairi- 

 ness in the race of L. diurna froin Sandford is 7-0284 - 4-4337 = •2-5947 units, or 

 allowing for the probable error of the start of the ränge, which is about -03 units, 

 52 + -16 hairs, the lowest ob.served number of haiis being 47. I feel böiind to call 

 attenti<3n to this feature of the distribiition, which is some sign Ihat in L. diurna 

 the ränge of Variation, in a normal hairy race, does not involve an occasional pro- 

 ductioii of glabious individuals; it is iuteresting in this connection to notice (1) that 

 Mr Bateson and Miss Saunders carefully refrain from speaking of a glabrous variety 

 of L. diurna, and (2) that the red-flowered hybrid, which Professor de Vries calls 

 Lychnis diurna glabra, results froni the cross between glabrous L. vespertina and 

 hairy L. diurna already alluded to. The lower limit to the distribution in L. 

 vespertina is 14-4399 — 13'63S2 = 0-8017 units, or 8 hairs per centimetre, which is 

 as good an approximation to a lower limit at no hairs as we need expect from the 

 limited number of observations. It must be remembcreil that the first unit of the 

 grouping adopted has a centre at 13 hairs, whilc the only individual in the group 

 was glabrous. 



The Cooper's Hill material secms heterogeneous, and I have not thouglit it 

 worth while to resolve this small number of individuals into componeuts. 



TABLE IV. 

 Number of Haiis with Glandulär Extremities per Hundred Hairs observed. 



* So my notes say. If we make tbe utmost allowance for possible brenkage of the tips of ßland- 

 hairg, so that they cannot be recogniseii, it is I think certiiin that no female plant from Cooper's Hill 

 had one per cent. of such hairs even near the base of tbe leaf. 



