TALTJE OF DIAGNOSTIC CHARACTERS. 



1 



such an extent that at length only a single character may be left whereby a diagnosi 

 between two, sometimes even more than two, species can be effected. Only a few of 

 the many examples of this need be mentioned:- P. Oederi can only be diagnosed from 

 P. flammea (unless seeds be available) by the hairs on its anterior filaments ; P. Clarkei 



from P. Pr aini ana 



(unless the colour of the flower be noted) by the presence of a beak 





to the corolla; P. corymbosa from P. crenata on the one hand by the absence of a beak 



and from P. zeylanica on the other by the much longer hood; P. rhmanthoid* from 

 P. siphonantha by its calyx being 5- not 3-toothed and by the apex of the beak of its corolla 

 being entire; while P. siphonantha itself is only witli certainty separable from P. lonqijhm 

 (if the colonr of the corolla is not noted) by the absence of hairs from in posterior 

 filaments and the presence of a distinct tooth on the margin of the hood. Careful 



analysis of the floral structure in every gathering is therefore absolutely essential for 



accurate dia 



5 



The opposite condition, with considerable difference of habit and great similarity of 



floral structure, is rare in this genus- Still instances do occur, and those of P. jn/cnantha 

 and P. Oederi, P. globifera and P. anas,, P. Gammicana and P. binaria, P. ( 1/rffH and 

 P. zeylanica, may be mentioned. The calyx affords less useful characters than the corolla 

 or the stamens, partly for the reason given above, partly because it is somewhat more 

 variable within specific limits. There do, however, occur instances where calycine 

 characters are of high value. One of these, pointed out by Sir J. D. Hooker, is the 

 case of P. fflwesii. Another excellent example is furnished by P. Pantlingii, which is at 

 once distinguished from the otherwise very similar P. fnrfuracea by its uncleft calyx. A 

 parallel instance among European species is the case of P. foliosa and P. sumana. 



The same thing is true of the gynsecium, less, however, because of the constancy of 

 its characters than because its modifications are usually accompanied by greater difl. rences 

 in the other parts of the flower. This remains largely true of the fruit. There are 

 indeed some instances where the capsule or the seeds afford excellent diagnostic 

 characters :— the seeds of P. flammea, as Dr. Lange has pointed out, are as distinguishable 

 from those of P. Oederi as are the stamens ; the characters of the seeds of P. Roylei make 

 it impossible to merge that plant in P. verticillata as Mr. Bentham proposes, whatever 



elasticity be given to the limits of the Linnean species ; and P. Wallichii differs even 



more markedly from P. asplenifolia, to which Mr. Bentham has referred it, by its 

 capsule and its seeds than by its habit and its flowers. Another case where the evidence 

 from the fruit is unmistakeable is P. longiftora, which cannot be confounded with P. 

 siphonantha at the same stage even if the habit, leaves, locality, date of collection, and 

 collector be identical, owing to a double curvature of the lower edge of its capsule which 

 the caosule of P. siphonantha wants, and to the absence of that pitting of the testa which 

 the seeds of P. siphonantha exhibit. But the general rule is the reverse of this ; where 

 there is identity of habit between two or more species accompamed by a greater or less 

 repetition of floral characters, the condition is even more marked in the capsule and the 

 speds than in the calyx. Among the most striking examples of this are P. carnosa, 

 P. corymbosa, and P. zeylamca, the capsules and seeds of which are practically indistin- 

 guishable ; P. pectinata, P. pyramtdata, and P. tenuirostris afford an equally instructive 



instance 



It is then sufficiently elear chat in this genus only floral characters can be relied on 



. ,. . i n;^ * twtna nnd reneated and careful analysis of the flowe 



in diagnosing nearly allied forms, ana iepe<ueu a j 



„ . • , fc« a ilia* annreciation of the limits of c>pecies even if 



suites of specimens is necessary for a just apprenai r 



