II. Vaiiatioii. and Cornlatlou in Lesser Celaiidine 147 



constaiits calcnlated by Miss Alice Leo. 'I'lu'so beioiiged to a variety of Le.sser 

 C'elaiuline clas.sed b}' tiie French biitauists as Fiairia calthaefulia and oonsidered 

 by them as distinct froin F. raimnculoidcs. Table I., however, e.xhibits iess 

 differoiitiatioii in the Horal parts of /''. cnltlKicßilia and F. runiiuculoides than can 

 occur betwcon two liwal rares of the lal,l,ei-. The Iudex Ketveiisis gives the forrner 

 nanio as a synonym of I he lalter, but there appears to bc a more sensible diifeien- 

 tiation of the leaf 



(B) 520 heads gatJiered on roadside banks at St Peter Port, Guernsey, 

 between April 2.")th and .SOth. Dne to Miss Caroline Herford. The characters 

 were counted, tlie tables fornied and Ihe Statistical eonstanls deterniined by 

 Miss Mary Reeton. 



(C) öOO heads gathered in two lanes and a nieadow near Thursley, on the 

 north side of Hind Head, Hurrey. 'J'he material was gathered between 12th and 

 17th of Ajiril by a nuniber of eollectors nnder the supervision of K. Pearson. 

 The characters were counted by Mr N. Blanchard and the tables and Statistical 

 constants are due to Dr Lee. 



(D) 50.5 heads gathered in a lane at Studland, Dorsetshiro, on April 7 by 

 Mr N. L. Blanchard. The counting is due to Miss Edna Lea-Smith and to 

 Mr Blanchard. The Statistical tables and constants are again due to Dr Lee. 



In reducing the material She])pard's corrections fir the moments were not 

 usöd, partly because we are dealing with Variation by units, and partly bccause 

 it is clear in the case of the sepals, and probably true for the petals, that the 

 frequency distribution has not high contact at the low end of the ränge. Other- 

 wise the ealculation of nieans, Standard deviations, coefficients of correlation, and 

 of Variation proceeded in the usual manner. 



(3) Table I. gives a summary of all the Statistical constants* for the four new 

 series and places alongside them those already found for othcr localities. Now 

 this table show.s at once the remarkable dififerences, which period in season and 

 environment can havc on mean, variability and correlation. Not one of these 

 quantities has the least approximation to constancy for all loeal races of Fiatriu, 

 nor for the same race at two parts of the same reason. The early Belgian 

 celandines, judging from the constants for stamens and pistils, which are all we 

 have, are practically identical with those from Italy, while the later ones diverge 

 widely and pcrhaps may be considered nearer to the Swiss Trogen series than to 

 any othcr. A brief study of this table will at once convince the reader of two 

 fundamental points : 



(f() Local races in plants cannofc be defined or distinguished by the existence 

 of differcnces many tinios the value of their probable errors between their mcans, 

 variations or correlations. 



(b) The influences of environment and season are for j)lants of siipreme inipor- 

 tance and very widely or indeed entirely screen any differences due to local race. 



* These are tabulated to four places of decimals — not because such ai-e exact or necessary forpresent 

 purposes, but because for future investigations when we come to consider the evohitionary history of local 

 races, it will be needful to have tlie constants to this number of places in oi-der to calculate the jiartial 

 regression coefficients true for each series to one or two places of decimals. 



19—2 



