C. H. OSTENFELD 119 



and produced a uniform offspring. It differs from the original form, 

 H. rigidum Alpha, in the following characters (compare the plates): 



1. It is lower and smaller in all its parts and the branches are more 

 divaricate. 



2. The involucral leaves (bracts) are more appressed than in Alpha, 

 where their tips are more or less recurved ; further they are of a lighter 

 colour (fewer dark hairs). 



3. Before the flowerhead opens the surfice is nearly flat, while in 

 Alpha it is distinctly concave, the corollas of the outer unopened flowers 

 reaching higher up than those of the inner ones. 



4. In the fully open flowerhead of the AlpJia-form the corollas of the 

 outermost flowers are much longer than those of the innoA- ones, while 

 in Beta the difference is not so marked and the diameter of the head 

 consequently somewhat less. 



I have shortly mentioned (1912, 1919) this result in some papers, 

 but as I had to deal with only one single case I felt it necessary to tiy 

 to get more material showing the same result. The appearance of the 

 deviating individual might be referable to some experimental error. The 

 probability for such an error was, as far as I could judge, only very slight, 

 as I did not know any form which was like the new one and as all the 

 seeds sown came from agamized heads, but on the other hand one indi- 

 vidual was rather unsatisfactory to build upon. Therefore I repeated 

 the whole experiment in an enlarged scale. 



In 1915 I agamized a good many heads of a plant of the original 

 H. rigidum and of another plant derived from the former after agamiza- 

 tion, thus both Alpha-individaah. The seeds of each of these two plants 

 were sown in 1916 and flowered in 1917. The offspring of H. rigidum 

 itself gave 635 plants like the parent and three " doubtful " ; the offspring 

 of the Fi of H. rigidum gave 469 typical plants and one "doubtful." The 

 " doubtful " plants were such as appeared to differ somewhat from the 

 typical, but as the plants stood rather densely, the differences might be 

 caused by conditions. Heads of the four " doubtful " plants and of one 

 typical were agamized, and the seeds were sown in 1918. The new plants 

 flowered in 1920, and it now appeared that the plants from three of the 

 " doubtful " parents did not differ from the typical. But the offspring of 

 the fourth " doubtful " plant (one from the seeds of the original H. 

 rigidum) gave 109 plants which were distinctly different from the type, 

 and also different from the first deviating form {Beta). We have thus 

 again obtained a deviation which was at once constant, gi^'ing uniform 

 offspring. 



