84 THE AMERICAN NATURALIST [VOL. XLIV 



vidual as the mother in one series of crosses, and a single 

 blue-flowered individual as the mother in another series. 

 The same blue-flowered and red-flowered plants were also 

 crossed at the same time with white-flowered plants. 

 The actual and theoretical results of these eight crosses 

 are given in Table I. 



In addition to these families which were bred in such 

 a way as to allow the definite working out of the gametic 

 formulae of the parents and the theoretical results, bluish- 

 flowered plants were also observed in a number of other 

 pedigrees. In some of these families only a small pro- 

 portion of the individuals had their tints recorded, as 

 they were being especially studied with other objects in 

 view. Such fragmentary records are of no special value 

 in this connection, of course, and they will not be pre- 

 sented, but in Table II. are given all those pedigrees in 

 which approximately all the purple-flowered offspring 

 were recorded either as "blue" or "red." 



In this second table it is impossible to vouch for the 

 correctness of the suggested theoretical results, as the 

 gametic formulae of the parents are in each case very im- 

 perfectly known. The column of theoretical results is 

 constructed simply by using that one of the available 

 theoretical ratios which fits most accurately the observed 

 facts. When numbers are so small, mere inspection can 

 not determine with certainty which is the correct theo- 

 retical ratio. Thus in No. 08168 the empirical ratio, 

 19 : 11 : 31, is almost equally well referred to either of the 

 available ratios, 1:1:2 and 3:1:4, as it stands about mid- 

 way between them. Notwithstanding the fact that igno- 

 rance of the gametic composition of the parents in this 

 second table makes it impossible to decide in all cases 

 what ratio should have been expected, the results har- 

 monize well throughout with those which comprise Table 

 L, where the theoretical "expectation" is definitely 

 known. 



All of the crosses recorded in these two tables seem to 

 be typically Mendelian, with the bluish-purple color hypo- 



