ASSOCIATION IN PAIRS 121 



Three conclusions may be drawn from this behaviour : (i) The 

 pairing properties of the chromosomes are specific to their constituent 

 particles (chromomeres), so that the whole chromosome does not 

 act as a unit, (ii) The affinities of the chromosomes are satisfied 

 by association in twos, (iii) Since the number of changes of partner 

 is limited it is evident that, as one would expect, an unpaired particle 

 is largely influenced in its choice of partner by the pairing of adjoin- 

 ing particles, the chromosomes pair in blocks. This, as we shall see 

 later, must have an important bearing on the conditions of pairing 

 in hybrids, which cannot be determined directly. 



In the pachytene thread, chiasmata arise as in diploids approxi- 

 mately at random. And since the chromosomes were paired by 

 blocks at random it follows that where enough chiasmata are 

 formed every possible combination of them is found amongst the 

 chromosomes associated. Amongst other combinations is that 

 where no chiasmata are formed at all between two particular 

 chromosomes where they have been paired, for sometimes with free 

 exchanges two chromosomes may happen to be paired for a very 

 short distance. Thus there arise, in triploids, single chromosomes 

 unassociated with their two homologues, and, in tetraploids, pairs 

 of chromosomes unassociated with their homologous pairs 

 (Hyacinthus, Primula sinensis). There can be little doubt from the 

 prophase observations that pairing has in these cases taken place 

 at pachytene between chromosomes which are left unassociated at 

 metaphase. It therefore appears that the association of chromo- 

 somes is preserved after pachytene and until metaphase solely by 

 the occurrence of exchanges of partner, chiasmata, amongst their 

 constituent chromatids (D., 1929 c). This conclusion is borne out 

 by quantitative observations of various kinds (v. Ch. XII). Its 

 significance will be considered later. 



Terminalisation occurs in the trivalents and quadrivalents that 

 arise from chiasma-formation at diplotene as in the corresponding 

 diploids. The associations produced in Primula sinensis (D., 1931 a ; 

 Dark, 1931) and Solanum Lycopersicum (Upcott, 1935) agree with 

 the assumption that every pair of chromosomes that have been 

 associated interstitially in one arm are at metaphase associated 

 terminally in that arm. When chiasmata joining three or four 



