68 Cooperatire Inrestigations on Plants 



low means have the four highest correlations and the four crops with high nieans 

 have the four lowest correlations. But even thiis the ortler of correlations is by 

 no means the order of relative means. We hold that, at present, all we can safely 

 concliide is that any treatment or change of environment of a crop which teuds to 

 raise its mean, will in this case lower its honiotyposis. Pcrhaps we niight even 

 say that 'starveling' conditions, which certaiiiiy existed in Chelsea I, O.xford and 

 Enfield II, teiid to intensify the homotypic relations, while favourable conditions 

 such as those of tlie magnificent Kidderminster crop, or the very good Crockham 

 crop tend to rcduce the homotyposis (see p. 58). 



The general average "498 of this series of eight agrecs excelleiitly with that 

 ■499 found for a former series of thirty-seven cases in the animal and vegetable 

 kingdoms* — only the first two crops being common to the two series. We see 

 that treatment may considerably influence iudividual results, but, as it appears as 

 likely to tend in oue as the other direction, the mean remains steady. 



(6) Methods of measuring Plant-character for Inheritance. When we are 

 considering quantitative inheritance in plants we are at once confronted by the 

 fact that allhough the flowcr, fruit, or Icaf may be convenient subjects for 

 Observation, they are not as a rule unique in tlie plant, and consequently we are 

 confronted with a nuilti|ilicity of elcinents for the deterniiiiation of the charactcr — 

 a very Utile Observation sliowing that variability within the iudividual is nearly as 

 marked as the variability of individuals among themselves. Thus a single fruit 

 or tlower cannot bc taken as a sample character of an individual, — for doing so is 

 at once assumiug that homotyposis is perfect or unity, wliich we know is very far 

 from the case. The consideration therefore of plant inheritance leads us at once 

 owing to the multiplicity of organs to homotypic relations. A like result rarely 

 occurs when wo deal with heredity in mammals, for their homotypes are not 

 Organs, which directly present themselves as suilable for the problem of inheritance. 

 In dealing with plants accordingly we cannot take a single organ, but must deal 

 either with all such organs, or at least with a sutKcieiitly large muddui sample of 

 them. 



If we express the character of the iudividual plant b}- the mean of any quantity 

 observed or measured on such organs, we are at once impressed with the varyiug 

 degree of weight to be given to this character according to the number of fruits 

 or flowei-s, etc. upon which the estimate is based. A Shirley poppy may have any 

 number of capsules from one to eighty and the average number of stigmatie bands 

 per capsule, which may be taken to express the individual plant character, will 

 have a great variety of weights according to the number of capsules dealt with. 

 In some plants it might be possible to take a random sample of 25 or 50 flowers 

 or fruits, etc. and strike the average of these. But this would be labiirious in the 

 case of a crop of 1000 individuals. ö to 10 as a sample would be easier, but not 

 nearly so reliable, and even so few could not have been obtained in the case of 

 many Shirley poppies. We therefore determined to take all the capsules and 



* Bivmelrika, Vol. i. p. 343. 



