3 
c) Plankton and nekton collecting apparatus 
d) Cephalopod collecting apparatus 
e) Surface temperature recording thermograph 
2) Hiring a marine ecologist to work on the ship, in con - 
,junction with the birdmen, or 
3) Train (i.e., provide schooling for) present field personnel 
for strictly marine ecological work. 
The above is not intended as a "model" approach to the situation but 
merely another call for help, and a strong vote for more simultaneous 
oceanographic data. 
C. Data Presentation 
That statistical approaches are the intuitive answers to such in¬ 
finite set problems as outlined in this introduction is sometimes too 
obvious. The statistics are too often taken for granted and can be over¬ 
worked and over-extrapolated. This sort of thing occurs repeatedly in 
simple cases of significant figures, i„e., reporting the percent of Bulwer 
Petrels in an area as 1.13 percent because three were seen in a total of 
264 birds. Even assuming the unlikely possibility that there is no bias, 
sample error, or gross mistake, and that there were in fact exactly three 
Bulwer Petrels in the delimited region, the representation of the fact by 
mathematical manipulation of a single digit and getting a 3 - higit figure 
is mathematical absurdity. 
The foregoing is by way of saying that the preservation of a large 
amount of data, i.e., the data reduction process, should be suited to the 
sample validity. In short, one can't make detailed conclusions, graphs, 
tables, etc., with crude data. Every effort should be made to adjust the 
data presentation so that statements, conclusions, and numerical sum¬ 
maries do not exceed the inherent vulnerabilities that come with all 
sample data. My course of presentation in this report has been to present 
the data in more than a single form and to assume more than one fixed 
standard of validity. This intentional shotgun technique is an attempt 
to present a choice of standards to be interpreted in as mucn depth, as xs 
deemed feasible by any reader. Even though many of the data in this re¬ 
port have better sample validity than many shorter surveys, I am aware of 
overstepping the bonds of statistical common sense in many cases. I have 
done this on data about which I have no good idea of the amount oi in¬ 
herent error, and, in some cases, on the grounds that it is important, to 
follow through with previously established format for the sake of com¬ 
parability. Where I have made general statements about qualitative (and 
quantitative) data I have attempted to make the statement unequivocable, 
self-definitive, and without hidden meaning, intended or implied. 
