xvi] Practical Hints 293 



seen a case of this kind. The true variety has a wrinkled 

 seed. The "rogues" had round seed. Every year the 

 variety was picked over by hand and the round seed rejected. 

 Nevertheless a pretty constant proportion of the rogues 

 persisted. Knowing that it was in the highest degree 

 unlikely that a wrinkled pea, being recessive in that respect, 

 could give off a round-seeded form, I examined the seed 

 with care and found that the seeds at the ends of the pods 

 of the rogues were liable to be shrivelled so much as to 

 pass for the true wrinkled type, and being thus admitted 

 into the selected seed, perpetuated the " rogues." This is 

 only one of many sources of error. Hitherto, through the 

 prevalence of incorrect views of the nature of variation and 

 heredity it has been thought more natural and likely that 

 plants should throw rogues than that they should not. 

 With the attainment of exact knowledge we see that the 

 opposite expectation is more probable, and in that hope all 

 operations of this sort should now be guided. 



Raising Novelties. 



When crossings between varieties are made, either with 

 the definite purpose of producing a combination of two 

 desirable qualities, or in the general expectation that some 

 novelty will turn up, it is scarcely necessary nowadays 

 to insist that the appearance and attributes of F^ the 

 first cross, give no indication as to the failure or success 

 of the attempt. That outstanding fact is at length very 

 generally known and appreciated, with the result that first 

 crosses are now preserved which a few years ago would 

 have been rejected. More important is it to lay stress 

 on the necessity for sowing a really large sample of the 

 seed from which /% is to be raised, for there must be 

 enough to give a chance of seeing the rarer combinations. 

 Since by the nature of the case most of the obvious crosses 

 between the familiar varieties of cultivated plants have been 

 tried in horticultural practice, the novelties are likely to be 

 found more often among these less frequent combinations. 



A Practical Example. 



As a good example of an F z family consisting of a 

 long series of types, that derived from one of the crosses 



