158 The Scottish Naturalist. 



Similarly with the genus Ranunculus. We find here a distinc- 

 tively generic mark, a mark that neither repeats the distinguishing 

 features of the Tribe, the Order, or the Cohort, nor trenches upon 

 the differentia that separate the various species of Ranunculi from 

 one another; and this generic mark would run: — "Stems hollow; 

 leaves sheathing at base, divided to a greater or less extent (except 

 in the Spearworts) ; sepals 5, sometimes (as in R. Ficaria) 3 ; petals 

 5, sometimes (as in R. Ficaria again) more, each with a hollow 

 spot (nectary or gland) near the base on inner surface, colour 

 yellow or limb of petal white or red ; carpels in a globular head." 



Now in all this I have aimed at three things : I have tried (1) 

 to include in a grade-mark only what ought to be included, (2) to 

 take in everything that may rightly claim a place, and (3) to indi- 

 cate degree of generality. And the full significance of what has 

 thus been done will best be seen, if, with a view to comparison, we 

 turn to any of the well-known botanical authorities, and note their 

 defects in method. These defects are precisely the ones that I 

 have attempted to avoid. Irrelevant characters are ever cropping 

 up, even in the best botanical works — characters that either have 

 not the degree of comprehensiveness that the particular grade 

 indicates, or that are repeated from some of the higher grades that 

 have gone before, or else characters in the form of mutually ex- 

 clusive alternatives ; little or no attempt is made to denote gene- 

 rality ; and lists of characters are given that are altogether inade- 

 quate and imperfect. Let us take an example from Hooker and 

 Bentham's " Genera Plantarum," for there is nothing like exempli- 

 fying from the best available sources ; and if the best, when weighed 

 in the balance, are found wanting, it is an argument a fortiori as 

 to the remainder. We begin with the tribe Ranunculese, whose 

 definition is given thus : — " Sepals imbricated ; carpels uniovulate ; 

 ovule erect, with ventral raphe; achenes indehiscent ; herbs ; leaves 

 radical or alternate." 



Now the first thing to be remarked about this is that we have 

 here both too much and too little. Too little; for the list is in- 

 complete, inasmuch as no notice is taken of the regularity of the 

 flowers (which is indeed a tribal character), or the tribal peculiari- 

 ties of the sepals and petals. Too much ; for characters are here 

 set down as tribal which have already been adduced as ordinal. 

 For, turn we to the order Ranunculacese, and what do we find? 

 We find inter alia that it is characterised by sepals imbricated, 

 achenes (when present) indehiscent, habit herbaceous, and leaves 

 radical or alternate; in other words three, at least, of the characters 

 adduced here as tribal are not tribal but ordinal. Moreover, 

 " indehiscent " achene is a tautology, inasmuch as the very mean- 

 ing ot the word achene is " a one-seeded indehiscent carpel." 



Regarded therefore from the standpoint of Method, this tribal 

 definition is vicious in the extreme, and, to any one trusting for 

 instruction simply to the book, misleading. 



