1914] Chandler: Fcaihcrs of Circus hudsonius 335 



feathers grow out in little bundles of from five to eight (pi. 

 16, fig. 2) on one side of the calamus of the contour feathers, 

 and they lie on the dorsal side of the feather to which they 

 belong. Invariably the filoplumes of a single bundle vary greatly 

 in length, ranging from 2 or 3 to 40 millimeters, usually no two 

 being of the same length. The longest ones found were attached 

 to the posterior row of under tail coverts. They are exceedingly 

 slender, being from 0.03 to 0.05 millimeter in diameter, and 

 entirely unbranched and unmodified up to the last few milli- 

 meters, which may simply bear a series of barbules, or may break 

 up into two or three barbs, each with its own barbules (pi. 16, 

 fig. 1). The barbules, which are widely spaced, there being eight 

 to fourteen per millimeter on each side, are entirely unmodified 

 and comparatively rigid, although filamentous and somewhat over 

 a millimeter long. The macroscopic appearance of these feathers 

 is like a miniature cat-tail reed, the long shaft, slender as a 

 spider's thread, corresponding to the stem, and the barbuled tip 

 representing the "cat-tail" itself. "What function can be served 

 by these feathers is quite inexplicable, as they appear too weak 

 and sparse to be of any possible use. 



As observed by Pycraft (1910), filoplumes are degenerate 

 feathers, each one, before development is complete, being sur- 

 rounded by a number of barbs. In Circus some of these barbs 

 appear never to spread apart and become fully diiferentiated as 

 in Pycraft 's figure, but they remain incompletely formed, em- 

 bracing the shaft of the filoplume. Pycraft intimates that this 

 shaft is merely a stronger persistent barb, comparable to the 

 barbs which ultimately disappear, but it should rather be looked 

 upon as a true shaft since barbs are never branched. It is 

 probable that the barbs which are lost belong to the shaft, but 

 through a defect in development never become attached to it. 

 While developing, the tuft of five to eight filoplumes with their 

 incomplete barbs appears like a single feather germ, but their 

 roots are entirely independent of each other. The filoplumes 

 do not lose their deciduous barbs until some time after the 

 contour feather has reached its full development and ceased to 

 grow. 



