GERANIUM PURPUBEUM 101 



reappeared, some of them with characters intermediate between it and 

 G. Robertianum. Presumably the hardiness and other characters of 

 G. Robertianum were dominant over the purpureum .characters in the 

 F 1 hybrid, the masked recessive pit rpi/rfiiim -characters reappearing in 

 1920 as the result of segregation. Possibly also some seeds had lain 

 dormant, but 1 feel rather doubtful whether they do so. 



A NEW EUROPEAN EPIPACTIS. 

 By Colonel M. J. Godpert, F.L.S. 



In 1868 Hermann Midler exhaustively described and figured 

 JEpipactis viridifiora Rchb. (Verh. N. H. Vereines der preuss. 

 Rheinlands). He stated that the stigma is turned so far back that 

 its upper half stands under the anther, that nothing is to be seen of 

 a rostellum, that the pollinia are deposited erect on their bases 

 on the front of the stigma, and that self-fertilisation is inevitable. 

 I was much puzzled to find that, in spite of these fundamental 

 differences, continental writers were almost unanimous in regarding 

 viridifiora as a mere variety of latifolia. Did they doubt Midler's 

 accuracy, or was his paper not widely known ? 



In June 1920 I went to Thorenc, above Grasse, where viridifiora 

 was said to be common, and to Mower, according to Camus, a month 

 earlier than latifolia. I was surprised to find its fioral structure 

 was exactly the same as that of latifolia. The stigma was in the 

 same position, the rostellum well developed, and the rapidity with 

 which the pollinia were removed from large numbers of flowers 

 showed that it was well visited by insects. Evidently this was not 

 Midler's plant. I sent specimens to Mdlle. Camus, asking if they 

 were really the E. latifolia sous-espece viridifiora of the Mom Orch. 

 de l'Europe, of which she was joint author with her father. She 

 replied that such was the case. Later Dr. Keller, of Aarau, kindly 

 sent me specimens of the Swiss viridifiora. In these the ros- 

 tellum was so effective that I found the greatest difficulty in opening 

 a bud without withdrawing the pollinia, through the viscid disc 

 adhering to the base of the epichile. Barla's figure of viridifiora 

 much resembles the Thorenc plant, and he shows the pollinia attached 

 to a well-developed rostellum. In all these cases the floral mechanism 

 is identical with that of latifolia. 



Among a number of the Thorenc viridifiora, gathered during the 

 previous three or four days, I found a single specimen of what I at 

 once saw must be Midler's plant. No rostellum was visible, and the 

 small pyramidal pollinia were standing erect on their bases on the 

 front of the stigma. I could not remember where I had found it, 

 and for the next fortnight searched feverishly in every direction for 

 further specimens, but in vain. At last, in a small pine-wood inter- 

 sected with paths quite close to the hotel, I found ten fine specimens, 

 and thought that I had now got the true viridifiora. From these I 

 compiled the following comparison of Midler's plant with latifolia 

 (including Camus' viridifiora). It should be carefully noted that 



