36 



revision in the Synopt. Flora shows that his work could hardly be 

 worse. A fairly successful move was the grouping into Sisymbrina 

 and Euarabis, but there it ends. The rest of the grouping rests on 

 stellate pubescence when there is none in the genus, on the vari- 

 ble earing at the base of the leaves, and on the arrangement of the 

 seeds in one or two rows, when everyone knows the seeds are al- 

 ways in two rows. If the pods are very narrow or about 1 mm. wide 

 the seeds mostly appear in one row, but as pods of the same plant 

 sometimes vary 3 mm. in width the seeds are in all degrees of va- 

 riance as to rows. This wholly unspeeific distinction is seized up- 

 on by Gnene and Nelson as a fruitful source of fictitious species. 

 Another very poor character adopted by Watson (and carried to 

 an absurdity !>v Greene and Nelson) is the development of ears. 

 Carrful field work shows that you can get almost any kind of ear- 

 ing in the same patch of plants; that perennials often bloom the 

 seco-.d yrar; that too little attention has been given to the endur- 

 ance of species; that too much has been made of the acuminatum 

 of the pod (ex< ept in the arcuata group), the same species varying 

 from »cnDiiiiate B to blunt ; that too little has been made of the size 

 of the seed.-; and thai rows of seeds and so-called "stellate pubes- 

 cence" are misUading. As a result of Watson's method we know 

 there is an incongruous mass of species on pages 161-2-3 (Synopt. 

 Fl.), while A. Breweri, which belongs with it, is far removed, 

 and A. platysperma with seeds decidedly in two rows is left in it 

 and far removed from suffrutescens, confinU Lyallii and Drum- 

 rnondii its nearest allies. A. Parishii and Cusickii are side by side 

 though not related. A. microphylla, canescens and Lemmoni are 

 much separated but closely related. A. Cusickii, Beckwithii, sub- 

 pn.natifida and lb.lbadlii are widely separated though closely re- 

 lated. A. pulchra and arcuata are widely separated but closely re- 

 lated. The chief character of pulchra is not noted, namely, that 

 its pods are cylindrical till mature, arcuata shows the same vary- 

 ing A. arcuata and Uolboellii are the most troublesome species 

 and though treated perhaps as well as can be done they are not 

 satisfactory. We have no evidence what the type of Holbcellii is, 

 can jud K e only b> geographical limitation. The looseness in de- 

 scribing species of Arabis by Watson is the chief cause of our 

 present .lirncmties. He always considered the Crucifera? a special- 

 ly -UMtural" family which was his way of saying that the distinc- 

 tion were artificial. In 1882 when I had carefully worked up in 

 the field the w.-siern Nevada forms he referred all my material to 

 old npecies (including my then unpublished A nulehra 1 * I re- 

 framed from publ,shi„g my A . pulchra on thetxpC ^understan- 



