LARKS. 485 



account of distantly related forms having been placed in juxtaposition in the linear 

 sequence. The allied forms must be looked for at the root of each series, not at the 

 end. Regarding our system as a tree, it is our plan to place it on paper by working 

 from the stem up the first branch to its top, then descending to the base of the next 

 branch, to again work upwards, and so on until the top of the last branch is reached. 

 In cases of reasonable doubt, however, we propose to give the conventional arrange- 

 ment the benefit of the doubt. 



Fi<;. ZX. Aldiula <irlirm, wood-lark (upper) ; A. arvensis, sky-lark (left); A, cristata, crested-lark (right). 



This last paragraph is applicable to the family of the larks, the ALAUDID.E. 

 They have sorely troubled systematists by apparently exhibiting characters of two 

 groups as diverse as the Motacillidre and the Emberizina3, and at the same time pos- 

 sessing a feature so unique within the present super-family as to prevent their proper 

 assignment to either. This peculiarity consists in the holaspidean tarsi, technically 

 making them scutelliplantar, the hind surface of the tarsus being broken up into scutes 

 similar to those covering the front part. This is generally regarded as a generalized 

 feature, hence the larks are usually placed at the bottom. We have a strong sus- 



