PROLIFERATION OF SPIKELETS IN BRITISH GRASSES 



47 



the time and place of collection. This in no way detracts from the value and importance 

 of Turesson's original work, but renders inadvisable the use of the key for wild specimens. 



The groups semiviviparae and viviparae are useful to denote partially and fully 

 viviparous plants. The two are easily determined, appear to be constant and have differing 

 potentialities. Both may be included in F. vivipara. 



4.4. Cytology and Origin 



All Turesson's seminiferous plants of F. ovina had a chromosome number of 2n = 14, 

 but the viviparae had numbers of 2n = 21, 28 and 42. Stahlin (1929) and Church (1936) 

 give 2n = 14 for " F. capillata Lam." ( = subsp. tenuifolia). They find or quote for various 

 seminiferous segregates of F. ovina agg. the following numbers : 2n = H, 2S, 42, 56 

 and 70. Thus polyploidy is not peculiar to the viviparous forms. 



Flovik (1938) investigated Spitzbergen plants and found : F. ovina var. brevifolia 

 2n = 28, F. rubra var. arenaria 2n = -42 i If and F. vivipara 2n = 49. Tromso 

 and Narvik plants gave : F. ovina 2n = 14, F. vivipara semiviviparae 2n = 21, F. vivi- 

 para viviparae 2n = 28 and F. rubra var. arenaria 2n = 42. On the basis of chromosome 

 morphology and numbers and the morphology of the glumes he suggested that F. vivipara 

 on Spitzbergen was the hybrid (2n = 49) of an unreduced F. ovina var. brevifolia gamete 

 (n = 28) and a reduced F. rubra var. arenaria (n = 21) gamete. In Norway he visualised 

 the reduced gametes of F. ovina (n = 7) and F. rubra (n = 21) giving rise to a fully 

 viviparous fescue (2n = 28), which by means of an occasional flower (see also Jenkin, 

 1922) backcrosses with F. ovina yielding the semiviviparous plant (2n = 21). 



Jenkin & Thomas (1949) report sickly seminiferous hybrids (2n = 42) of F. ovina 

 and F. rubra. Nannfeldt criticises Flovik's hybridisation schemes, because the general 

 form and growth habit of F. vivipara is not intermediate between the supposed parents. 



Seed was obtained by Jenkin (1922) from late-arising inflorescences of viviparous 

 fescues, pollination being possibly by nearby seminiferous Festucae. Four plants were 

 raised from seed, two being virtually normal seminiferous plants. The other pair showed 

 strong signs of the parental vivipary. One would appear to have been a fairly normal 

 semiviviparae form; the second was somewhat monstrous. 



The occurrence of flower-bearing by viviparae and the existence of the semiviviparae 

 permit of genetic exchange between the viviparous forms and their seminiferous relatives. 

 Thus, even if vivipary originated among the fescues but once, there is no reason to expect 

 uniformity to have remained. The origin of F. vivipara is unknown, but its variation is 

 certain. 



5. POA BULBOSA VAR. VIVIPARA KoEL. 



Caspar Bauhin (1620, 6) described Gramen arvense panicula crispa and illustrated it 

 by a woodcut. Tournefort (1700, 1, 523) quoted this, but renamed it as Gramen panicula- 

 tum, proliferum. Scheuchzer (1719, 45) cited both the former authors and Jean Bauhin, 

 and included the plant in his group of proliferated grasses. It is difficult to tell if Bauhin 

 meant only the viviparous form, although the already mentioned error by Ray indicated 

 that he believed this plant to be viviparous. Bauhin's plant was certainly of Poa bulbosa 

 agg. and a seminiferous form was separately described by Scheuchzer (1719, 40) under 

 Lobel's name Gramen xerampelinum. 



The specific name Poa bulbosa was due to Linnaeus (1753, 1, 70) and all the above 

 names are quoted. Halperin (1933) has reviewed the subsequent literature, stating that 

 all except one of Linnaeus' specimens were viviparous. Halperin concluded that the 

 correct name is Poa bulbosa var. vivipara Koel. (Koeler, 1802, 189). 



Figure 5 illustrates spikelets of both forms. 



