
cards are correctly punched (and punching error averaged about 15 per 
thousand in the cards I used), the electric contact brushes on sorting 
machines are subject to inevitable wear. In the early stages of wear, 
an occasional card will be missorted as a result. I routinely thumbed 
through machine-sorted cards to catch errors in punching and to learn 
when the contact brush on the machine should be replaced. In compiling 
life tables for each species except mallards, I routinely sorted the 
cards twice. 
A third source of error seemed to lie in my own assembling of 
usable recoveries. Reports of bands “found on a road" or "in the school 
yard" obviously are of little use in life-table work. I deemed it also 
advisable to discard reports of "skeleton found" and "remains found." 
(In small passerine species, these are unimportant; in pelicans they 
would tend to be a source of bias.) When specific methods of recoveries 
were being tabulated, reports giving "no information," "taken," "picked 
up," and "found" were also eliminated. All such discards in the white 
pelican amounted to 12.1 per cent of the cards on file. Together with 
conflicting or vague reports and birds found dead in colonies, they 
represented 20.2 per cent of the double-crested cormorant file. Since 
these eliminations were made without reference to date of recovery, I 
do not believe they introduced any new bias into the samples studied. 
Their removal, however, was often carried out by thumbing and visual 
inspection of each card. Some very thin cards tended to stick together 
so that in very large samples a small sorting error seemed to persist 
in spite of much precaution. 
Variations in the manufacture of bands.--The bands placed on 
wild birds have varied considerably Indesign. ‘Examples of those issued 
by the American Bird Banding Association during the second decade of 
this century were examined by me at the Patuxent Research Refuge. 
These are extremely thin bands and are not nearly as thick as those 
currently issued by the Fish and Wildlife Service. In my own experience 
with herring gull bands during the 1930's, when the manufacture of the 
bands was reasonably standardized, perceptible annual differences in 
the brittleness of the modern bands were also evident. More recently, 
Austin (1947a) has found tern bands mich softer than usual and substi- 
tution necessary after two years. F.C. Lincoln (personal communica- 
tion) informs me that variations in the composition of American bird 
bands have undoubtedly persisted since the program was taken over 
by the Federal Government; he believes that particularly fragile bands 
were used up to 1925. In nearly all cases I used this as the starting 
date for samples that I analyzed. 
Environmental variations in wear.--The widely differing habits 
of the various families of birds subject bird bands to equally wide varia- 
tions in wear. Severity of use theoretically progresses from air-dwell- 
ing species (swifts and swallows) and through the tree dwellers (vireos 
and kinglets), weed and grass dwellers (quails and pheasants), fresh- 
water swimmers (river ducks), and salt-water species (eiders and scoters). 
The severest use seems to be that associated with salt-water-feeding, 
rocky~ledge-nesting Alcidae and burrowing Procellariiformes (Lockley 
1942, pp. 117-118). 
23 
