Some Questions which they Suggest. 35 



and observe its stalk, we shall find that it is four-sided, so 

 that a section across it is a square. Now this characteristic 

 might easily be supposed to be one of little consequence, 

 and yet, in fact, it will be found to be a true and valuable 

 one, and that all plants with a square stalk and lipped 

 flowers will be found to have a four-lobed ovary and four 

 nuts on the bottom of the calyx, and these belong to the 

 family of the Labiatse. If now, on the other hand, we 

 count the number of the stamens in plants, and use this 

 character as the foundation of our classes, we shall break 

 up this natural family with its square stems, and shall 

 relegate some genera, such as Salvia, to one class, while the 

 great mass of the family go to another, and, what is 

 perhaps worse, these exiled genera find themselves put 

 into a class together with plants with which they have no 

 real connection or sympathy with the Enchanter's 

 Nightshade and the Duck- weed. This form of the stem 

 then has a high value as co-existent with a general likeness 

 of structure ; the number of the stamens may vary in 

 plants closely akin, and agree in plants widely different, 

 and therefore has a low systematic value. 



The variations of form of our domesticated dogs are 

 generally held to be of no value even as specific distinctions ; 

 but the difference of the markings in the spores of myxies 

 is held by those who have most studied their classification 

 to be often a safe difference as between two species. It is 

 only by experience that we can tell the systematic value 

 of a difference i.e., by observing how far it is correlated 

 with other differences of structure or life-history, and 



D2 



