catagories 
The Reliability Classes set forth in C.L.# 4 do not cover the 
reliability ranges adaquately, in my opinion. The three classes 
(at 10$, 50$, and a guess) give too much of a range. Most of our 
data here on Sand, for example, could not fall in any of these three 
classes. This mainly applies to those groups which are §# present in 
small numbers. There is a big difference in 10$ and 50$ and a guess. 
Irksome cases we know exactly how many birds are present on Sand Island 
(i.e.,Blue-faced Booby.). This applies more so when giving numbers of 
nests and young. I suggest that the classes be extended to include: 
A. Direct count 
B. Believed to be accurate within a £ 10$ error 
C. Believed to be accurate within a £ 25$ error 
D. Believed to be accurate within a sfc 50$ error 
E. Essentially a guess 
I have used these five classes in the hiarch 16-3^ report. 
As I said earlier, I think that BI-WEEKLY BIRD SUMMARY REPORT 
as presented in C.L.#4 is exactly what we need for reporting out bird 
surveys. I hope a final form will be decided on in the very near future 
so that forms can be printed for field use. I will try to get some 
ditto copies for our immediate use here on Sand f but the 31 should 
standardize the reporting form. 
Best regards, 
Binion Amerson 
Research Curator, Pacific Project 
Division of Birds 
