212 
Part III. — Tenth Annual Eeport 
hypothetical one, in reference to fish food. Suppose that 100 cod were 
examined and that 25 of these were found to contain annelids. By 
Poisson's Rule the degree of the limits of error may be ascertained. Let 
M = total number of cases recorded in the series of observations (100); 
m = the number of cases in one group (25); n = ihe number of cases in 
the other group (75). Then m + n = M i and ^ and are the propor- 
tions of each group to the whole. But in another series of 100 obser- 
vations, made say in similar circumstances, the proportion may be 
That is to say, working out this in values, that in another series of 100 
observations the number of cod containing annelids instead of being 25 
may be as many as 37, or as few as 13 — a variation of nearly 50 per cent, 
of the actual number found. If, however, instead of 100 cod having 
been examined, there had been 10,000 examined, the limit of error would 
have been only a little more than 1 per cent, above or below the propor- 
tion actually found. From this it is evident that observations made on a 
very small number of fish are of little value, so far as regards general 
conclusions from the observations. This fact I have kept constantly in 
mind, and in recording small numbers of fish examined, I have forborne 
to draw any conclusions whatever. Without considering fully the subject 
of time and space distribution of food, I would give here some of the 
collective results of the four years' investigations of, What do fish eat 1 
FIRTH OF FORTH. 
Plaice. 
In all 1205 were examined in the Firth of Forth, of these 883 contained 
food material that could be identified. 
Echinoderms were found in 89 stomachs (10 per cent.). They consisted 
almost entirely of sand-stars. The following is the list of forms found ; 
Ophioglypha in 21 ; Amphidotus in 7 ; Amphiura filiformis and other 
species in 27 ; Echinocyamus in 2 ; and Ophiura albida, Spatangus, and 
Echinocardium, each in 1. The rest of the echinoderms were unidenti- 
fied sand-stars. 
Annelids were found in 485 stomachs (54 per cent.) In the case of 
362 stomachs they were not identified. The identified forms found were 
Sabella in 35 ; Aphrodite in 25 ; Priapulus, in 20 ; sp. caudatus in 9 ; 
Nereis in 11; Sipunculus in 10; Arenicola in 5; planarians in 5; 
Echiurus in 3. 
Arthropods were found in 43 stomachs (4 per cent.). The forms found 
were Crangon in 7 ; Eupagurus in 7 ; Ampelisca in 6 ; Povtunus in 6 ; 
amphipods in 3 ; Psamatlie in 3 ; Porcellana sp. longicornis in 2 ; Pand- 
alus, Hyas sp. coarctatus, each in 1 ; unidentified hermit-crabs, and 
unidentified crabs, each in 3. 
Molluscs were found in 326 stomachs (37 percent.). The forms found 
were Scrobicularia in 208 ; Solen in 51 ; Mactra sp. subtruncata and ? in 
13; Venus sp. lincta, fasciata, and falnia in 8; Pecten sp. tigrinus, 
siriatus and opercularia in 8 ; Cardium sp. ecliinatum and 1 in 6 ; Nucula 
sp. nitida and 1 in 5; Psammobia in 4; Astarte sp. compressa in 2; 
Buccinum sp. undatum and ? in 2 ; Cyprina, Corhda, Leda, Tapes vir- 
ginica, each in 1 ; unidentified lamellibranchs in 22 ; unidentified gas- 
tropods in 1. 
