ALIMENTARY TRACT OF CULEX PIPIENS 613 



since the homologous members of each pair show a tendency to 

 fuse and form a single bivalent chromosome, we should expect 

 to find the chromosomes of the multiple complexes also con- 

 jugating, not in three pairs, but in three groups of homologous 

 individuals. It is beheved that such is really the case. 



We find the fully condensed chromosomes still arranged in 

 groups of three, the difference in size of the three kinds making 

 it easy to see that we have sister chromosomes in each group. 

 Compare figures 21 A, B, C and D for relative .size of the three 

 chromosomes in normal complexes. But the individuals of a 

 group do not always divide at the same time, which accounts 

 for the finding of nine, eighteen, and thirty-six chromosomes in 

 some cells, whereas, simultaneous divisions of all the members 

 of a complex would give twelve, twenty-four, etc. It is interesting 

 to note that when an early division takes place in one of the mem- 

 bers of a given group, the same thing occurs in the corresponding 

 member of each of the other two groups. A good many com- 

 plexes have been found in which this point is brought out very 

 clearly through inequality in thickness of the individual threads. 

 In figure 17, A and B are evidently sister chromosomes, possibly 

 derived from the mate of E ; C and D show the same relation to 

 F. Figure 19 brings this out even better. A, B, and C evi- 

 dently represent one pair of which one member has divided 

 into two, A and B. Figure 20 illustrates the same point. 



It is obvious, since every group counted belongs either to a 

 'six series' or a 'nine,' i.e., six, twelve, twenty-four, forty-eight, 

 or nine, eighteen, thirty-six, seventy-two, etc., that, when 

 irregularity in the time of division of members within a group does 

 occur, it must come with the first sphttin'g of half of the threads, 

 and that otherwise all the derivatives of each of the daughter 

 threads of the first spht must divide simultaneously. If this 

 were not the case, we should get multiples of numbers other 

 than six and nine. We might suppose an early division of either 

 the maternal or paternal element of each pair to have given us 

 the nine chromosomes which we find in some complexes. The 

 fact that in some individuals we find only the 'nine series' — 

 nine, eighteen, thirty-six, could be explained by supposing the 



