8 July, 1908.] Improvement of Cereals by Selection and Crossing. 409, 



the proportion of red to white in subsequent generations is similar to that 

 of tall and dwarf, as shown in the following scheme : — • 



Parents— Red X White ^Howered) 



I 

 1st Generation — all Red 



2nd Generation — Red (pure) Red Red White (pure). 



3rJ (jeneration— Red (pure) Red Red Red White Red Red Red White Wliite (pure) 



There is here the proportion of one pure red to two hybrid reds and one 

 pure white. That this will be roughly the proportion is seen by taking counters^, 

 such as red and white slips of paper, to represent the red and white- 

 flowered respectively, and placing 50 red and 50 white in one hat and the 

 same number in another hat. By drawing one counter at random from each hat 

 and arranging the pairs according as they are red and red, red and white, 

 or white and white, it is found that they approximate to this proportion.. 

 In my test with counters, in the first fifty pairs there were 12 RK-, 22 R\V 

 and 16 WW, and in the second fifty pairs there were 14 RR, 24 RW, and 

 12 WW. The total proportions were 26 RR, 46 RW and 28 WW and the 

 larger the number of trials, the closer will the proportion approximate 25, 

 RR (pure reds) 50 RW and 25 WW^ (pure whites). That the germ or sexual 

 cells of the hybrid carry the unit characters or initidls, as they are some- 

 times called, in a pure and not blended condition — that they are segre- 

 gated or sorted out in the pollen-grains and egg-cells — has been proved by 

 numerous experiments and in this simple fact lies the key to the e«planation 

 of the varied phenomena of crossing. 



From this conception of unit-characters the individual must now be 

 regarded from a different stand-point. Instead of regarding the individual 

 plant as a unit, we have now to conceive it as made up of separate inde- 

 pendent characters and the aggregate of these constitutes the individual. 

 The individual is thus a living mosaic and if we removed all the separable 

 cliaracters we would in all probability come to the v;mishing point. 



This conception of distinct unit-characters is similar to <^he assumption 

 of atoms in chemistry, and like atoms they are regarded as indivisible 

 and they serve as a tangible basis for showing what happens when hybrids 

 are produced. We have alreadv given an illustration of what happens 

 when tall and dwarf characters and red and white-flowerino; are blended 

 in peas. The cross-fertilized plant produces its seeds as usual and when 

 these seeds are sown the progeny is not quite what one -.vould naturally 

 expect. In the examples given, instead of being intermediate in height 

 or light-red in flower, they are all tall and all red and when the seeds of 

 the first generation are sown again, the result is a mixture of tall and dwarf 

 or red and white in certain proportions. All the possible forms have now 

 appeared in the second h\ brid generation, which is often called the vari- 

 able generation, and a certain proportion of these forms is already fixed. 

 If the breeder saves the seed from each plant separately he will be able 

 to indicate the fixed individuals. 



There is thus a great contrast between the first and the second pro- 

 ducts of crossing. The first products may have characters intermediate 

 between the two parents, or they mav resemble one more than the other, 

 but whatever the change is all the seeds have it alike, so that the products 

 appear exactly alike. But the second products, instead of being moulded 



