410 
Cooperative Investigations on Plants 
susceptible of much alteration. Lastly, although we have classified each character 
into more sub-groups than we have met with in Mendelian writings, we find many 
cases in which the category is hard to determine, and we are convinced that 
personal equation must be given its due weight. 
(6) Parental Correlation. 
As we stated at the beginning of this paper, we intended to base the 1903 
experiments on examinations of apical flowers only. We were not therefore able 
to compare in the bulk of characters the crops of 1903 with the maternal plants. 
Ample material should, however, have been provided for parental correlation in the 
three crops of 1904. Unfortunately two of our observers were unable to complete 
their tasks. We are therefore thrown back on a comparison of the Crewe daughter 
plants with the Parkstone maternal plants. The fact that the Parkstone and 
Crewe crops were grown in very different environments is against very good results, 
but the number of bagged capsules at Parkstone and their comparatively good 
state led to our selecting that crop as the chief source for the seed supply of 1903. 
The tables are formed for the apical flowers of the maternal plants at Parkstone 
and the daughter plants at Crewe, and this was done with six characters. The 
work of tabling and reduction is due to Dr A. Lee. The coefiicients of correlation, 
except in the case of the stigmata where the method of moments was used, were 
found by the fourfold table method. We should, if the investigations were now to 
be started de novo, probably use the method of mean square contingency, but it 
was considered desirable to preserve as far as possible uniformity of treatment 
throughout the whole round of poppy investigations, and the fourfold table method 
had been adopted in the first memoir. We refer briefly to the characters dealt 
with. 
Stigmata. There was probably a stringency of selection here of the maternal 
plants of about "7295. It is an open question to what extent the homotyposis of 
about "5 ought also to be allowed for, at any rate the value found for parental 
inheritance is a minimum value. 
Extent of Margin. This was determined by a fourfold table, the divisions 
being between ' no ' and ' some ' margins. The factor of selection of the maternal 
plants is here "6259. The homotypic influence is, as in the case of colour, 
probably very small, the flowers on a plant being usually all whole coloured or 
all margined. 
Nnmher of Petals. This was determined by a fourfold table, division into 
normal four petals and non-normal flowers. There was inverse selection of 
maternal plants conopared to the general Parkstone population, i.e. they were 
more variable than that population ; but if we compare the maternal group with 
the Streatham crop as a Shirley population of maximum petal variability there 
would have been a stringency of selection of "5278. It is not clear, however, 
how great a selection we ought to allow for, because the observers may not 
have been equally definite as to what they reckoned as an incipient petaloid 
