92 THE COTTON PLANT IN EGYPT chap. 



Whenever a family is henceforth meationed, it should 

 be understood to be one of these strains. Commercial 

 " varieties " wiU be specifically designated as such. 



Since the fluctuation-graphs for parents will necessarily 

 be consulted when dealing later with genetics proper, the 

 reader is referred to these, except where some special 

 point requires to be illustrated by a special figure. 



For the expression of fluctuation in statistical terms, we 

 shall take the percentage probable error, and not the 

 " coeflficient of variation." The latter is the " standard 

 deviation" expressed in percentage of the mean. The former 



is roughly two -thirds of the latter, i.e., PE = 0'67\/ -, 



and our particular expression is — ^^r^= . This form is 



convenient, since half the observed cases must lie within 

 the limits given. The maximum possible true fluctuation 

 may be taken as odds of about 30 : 1, or 3*2 times the 

 probable error. 



Whenever a specific figure is given for a family it 

 should be understood as having been checked by com- 

 parison with other families, and therefore considered by 

 the author as being fairly representative of the data 

 available. 



Colour characters. — Such characters as the colour of 

 the petal anthers are devoid of appreciable fluctuation. 

 This uniformity is probably apparent rather than real, 

 since a diff"erence of less than 10 per cent, in the 

 intensity of a colour cannot be perceived by the ordinary 

 eye. 



When colours have to be classified, as in hybrid popula- 

 tions, standard flowers, &c. are always used, and the 

 unknowns are matched to them. 



Characters which depend on the presence of antho- 

 cyanin in the tissues, such as the red spot at the junction of 

 petiole with lamina, fluctuate widely. Comparisons can 



