!.!.\- I!"N-IVIKNMMK> \\III1 KM 1! A'. KM AMI lil.\i.INI 



17:; 



the sources of fallacy wing the number and 



changing the concentration of the reagent* and in 

 m;; tin- -Mnil.inl of \ulu v .th the abscis- 



sa; here u-ed. Notwith-t.iinliii;; the crudities of the 

 in< ti: '- a<! ;' I ami the fallacies introduce. I in the 



f.innulati f the composite charts in the former 



Mowing wu rendered apparent: That the 



f members of a genus constitute a well-defined 



. the mean f the character-values constituting a 



dMinrt L-i-iierii- t\ . j>e tending to be similar to 



the t\ ]>: of very closely related genera and dissimilar 



to the types of dixtantl'y related or unrelated genera; 



that the r.-a. tions of different species of a geniu yield 



HIM' nd to be closely in conformity with the 



generic type of rune, hut when there are representative* 



1 'genera or similar _. M.TI.- subdivisions there may 



ires or aberrati"iis from tin- generic type so 



that there may be as many subgeneric or group type* as 



nera or sul>geiieric groups; that the reac- 



>4ofasp> i curves that very closely 



1 with thnae of the species ; and that the generic, 



. and species differentiations arc in general 



e accord with established botanical data. The rc- 



>f the present research are in harmony with those 

 nf tin- prei-cdini: investigation, but some unexpected 

 variations have been found, especially in the extent of 



i'Tie and subgeneric dilTerentiations which will 



I to here with sufficient detail. 



Taking up first those genera which arc bent repre- 



! by -|XM j.-s ainl varieties, but in which there are 



not inrliideil Mih-.'cneric or similar generic group rcpre- 



!i a.< ll\i>i*atrvm (Charts K 2, E 3, and 



. Chart> K 10. K 11, and E 18), Narcissus 



-.'I. inclusive), and Lilium (Charts 



I, inclusive), it will be apparent UJMUI 



-u|HTticial examination that the starches of the 



varieties or species, or of both varieties and species, of 



each genus have curves that are in general very similar 



in form and that the type form of the curve in each genus 



t from that of any other, and so markedly 



so that the curves of the members of one genus could 



not he confounded with those of another any more than 



could the plants themselves. It will also be noted that 



when the starches are from very closely related plants, 



as in the Ilipiwastrumx, the curves arc very closely alike, 



while in \erinf and .Vomwus, respectively, where there 



are instances of both botanical closeness and separation, 



the variation* from the mean or the generic type of 



tend to be more and more marked as the repn- 



: ives of the genus are botanically farther separated. 



The curves of Lilium. while yielding a generic type very 



different from the //i'/i/w.frum, fferine, and .YarrtMtM 



types, arc of little usefulness in the differentiation of 



.arious member? of the genus represented because 



very rapid gelatinization of the starches with 

 nearly all of the reagents. In order to satisfactorily 

 differentiate th.-e starches reagents of such modified 



:ui*t be used as will render gelatinization very 

 much less rapid, and probably additional reagents may 



'-;;iry. In <>thT genera studied, where there are 

 nnly the two parental and the hybrid representative* of 



_renu*, as in Qlatliolus (('hart K31). Trilnnia 

 On (Chart E40). MUM (Chart 

 Kill. /V...V fihart Miltonia (Chart I 



Cymbidium (' ^ponding peculiarities 



will l>c found, although in HlnJiolus ami Trilonia. closely 

 related gem m, tin- curves are so much alike a* to indi- 

 rate different species rather than different genera. There 

 is also much resemblance between th>> Amaryllis and 

 1'haiiu charts which represent very widely separated 

 genera, but this singular peculiarity will If rvferr 

 particularly later on. In the Amaryllin-Hrunfvigia reac- 

 tions (Chart El), where there is bigcncric representa- 

 tion, the curves are quite different . 



When genera are represented by subgcnera or cub- 

 ic groups, as in Iliiinnnthus (Chart K < I. Cn'num 

 (Charts E7, E8, and E9), Iris (Charts E 30, E31, 

 and E 33), and Begonia (Chart E 30), the curves 

 of the subgeneric representatives may differ not only 

 markedly but to even a much more marked degree than 

 the curves of different genera generally of the same 

 family a most curious and n yet inexplicable phe- 

 nomenon. In llirmanlhu* the curve of //. puniftut is so 

 variant in comparison with those of //. kallirrintr. II. 

 magnifinu, and both hybrids that it seems that this spe- 

 cies must be separated botanically sufficiently far from 

 the other two to be regarded a* bclon^in^ to a different 

 subgenus, although this differentiation may not have IHI-H 

 recognized by the systematist. In Crinum the curves of 

 the representatives of the hardy and tender forms (C. 

 moorrt and C. longifolium, hardy ; C. ;ri/lnni' inn. tender ) 

 differ so markedly as to suggest mcmlxTs of different 

 genera. In Iris, in the first three sets (Charts K :?". 

 E 31, and K 32), the reactions of rhyzomatou* form- are 

 represented, and it will be Keen that all of the curves 

 conform closely to a common type; but in the fourth set 

 (Chart E33) the reactions are of tultcrous forms, all 

 three curves conform with great closeness to a common 

 type, and they all differ materially from the rhyzomatous 

 type, and in fact so different are they that they would 

 certainly not in the present stages of the investigation 

 be recognized as belonging to the same genus. In llr- 

 gonia there is found an even more remarkable instance 

 of subgeneric differentiation in the curves of the tuU-rou. 

 and semituherous forms, the former l>eing repre*' 

 by four garden varieties and the latter hy //. socotrana, 

 a very exceptional and isolated species of the genus. 

 Comparing the curves of these charts (Charts E 36 to 

 K3!) it will be seen that the curve- of the tuberous 

 forms are in close conformity to a common type, while 

 the curve of B. socolrana. is so very unlike the curves of 

 the former in a large number of the reactions with the 

 chemical reagents as to suggest anything but generic 

 relationship to the tuln-nms forms, rnfortunately. the 

 number of reactions of the latter were with a single ex- 

 ception very limited, hut the curve of the reactions of B. 

 tingle crimton tear! ft (Chart E30) can with perfect 

 safety be taken as very closely typifying the curves of 

 the others. 



The Amarytlix and Phaitut curves (Charts El and 

 E42), while representing wholly unrelated and widely 

 separated genera, give the impression of curves of closely 

 relsted genera or even of species of a genus; in far* 

 reaemblance is much closer than that of related crenera 

 here represented, as, for instance, of A marylli* and Brunt- 

 rigia (Chart E 1). "f t'haiu* and Miltonia (Chart* 

 and K ID. or of PAoiiw and CymbMium (Chart* 

 and K II). While there is some resemblance l-tween 



