Fishes Killed by 1950 Eruption — Gosline et al. 
25 
tently their scales would fall off. Neverthe- 
less, the specimens in these four collections 
are probably in a better state of preservation 
than is most deep-water fish material. 
Certain highly desirable data concerning 
our collections are unobtainable. For one 
thing, little can definitely be said about the 
depth from which these fishes came. Some of 
our specimens, such as the brotulids, are deep- 
water bottom forms. Others like the mycto- 
phids are bathypelagic. A few probably 
normally inhabit the surface layers of the open 
ocean. A very few are obviously inshore forms 
which had drifted into the area where we were 
collecting. The maximum depth at which our 
fishes were killed is likewise unknown. The 
most that can be said is that the known depth 
records for such fishes as Opisthoproctus , of 
which we took five specimens, indicate that 
at least some of our material is from rather 
deep water, presumably greater than half a 
mile. 
We also do not know how representative 
our collections are of the total fish fauna 
living in the area. Four types of selectivity 
have probably restricted the representative- 
ness of our sample. Two of these are due to 
our collecting methods and can be taken into 
account. First, we made no effort to collect 
the obviously inshore forms. Second, small 
fishes were almost completely missed because 
we took only what we could see and catch 
with a dip net from a boat. The other possible 
selective factors are: differential decimation 
Fig. 3. The third lava flow entering the water several days later. Photograph by Jack Matsumoto, 
