THE FLOWEKS OF TBAGOPOGON 255 



conditions. This fact, not very generally recognized, was noted by 

 Lubbock nearly fifty years ago {British Wild Flowers in relation to 

 Insects, p. 22, n. ; 1875). 



(3) That, roughly speaking, its flowers remain open for no more 

 than about four hours out of the twenty-four. 1 observed also — 



(4) That its flowers are not ephemeral : that is to say, each 

 individual flower opens on more than one day; but on-how many days 

 in succession it does this I failed to ascertain. 



Later, I made similar observations, as stated alreadjr, on the 

 closely-allied Tragop>ogon porrifolius — a plant 1 have been accus- 

 tomed to grow in my garden for culinary purposes. As a rule, the 

 plant when grown with that object is not allowed to flower, being 

 dug up to make room for some new crop as soon as its roots are ready 

 for the pot. Consequently, few people are aware how fine a sight a 

 bed of Salsify in full flower may present, during the following 

 summer, if surplus roots, instead of being dug up and thrown away, 

 are allowed to remain in the ground and produce blossoms. In that 

 case, its handsome purple flowers will present, early in the following 

 June, a gorgeous mass of colour which will astonish those who know 

 the plant merely as a pot-herb or as a casual escape from cultivation. 

 Further, after its flowers have gone off, its seeds (each of which is 

 provided with a " parachute," exactly like that of T. pratense) will 

 attract crowds of small birds (especially, in my case, Goldfinches) 

 which devour them greedily. 



The Mowers of T. porrifolius resemble in all respects except colour 

 those of T. pratense. In respect of opening and closing, they behave 

 in exactly similar ways, except that they appear both to open and 

 close slightly earlier in the day. Thus I observed many times that 

 they opened, as a rule, soon after 7 a.m. and closed well before noon. 

 Their open-hours were, however, dependent to some extent on the 

 prevailing weather conditions, as in the case of T. pratense. Thus on 

 10th June, 1916, when I had two separate beds in full bloom, most 

 of the flowers did not open at all and the rest to a small extent only, 

 the day being stormy, dark, and overcast. 



The autonomous openings and closings of flowers in those species 

 in which the open-hours are limited to a few special hours during the 

 twenty-four (the particular hours varying according to the species) 

 have never been, I think, adequately studied. 



One's first thought is, naturally, that the open hours of any 

 particular species of flower will be found to synchronize with the 

 hours of flight of the particular species of insects which habitually 

 pollinate it, just as some flowers which are regularly pollinated by 

 night-flying moths remain closed during the day and open only in the 

 evening. This theory is plausible enough in itself ; but in the case 

 of Tragopogon it seems entirely inapplicable ; for the insects by 

 which it is usually visited (namely, Dipteia, Hymenoptera, and 

 Lepidoptera : see Knuth, Pollination of Plants, ii. 679; 1908) 

 remain habitually upon the wing throughout the whole of the day. 



Another very plausible theory is that plants which are normally 

 pollinated by large day-flying insects may, by opening and closing 

 early in the morning only, escape the depredations of small, creeping, 



