136 
BULLETIN OF THE UNITED STATES FISH COMMISSION. 
tlian under entirely natural conditions. We would expect the death of the male 
and the introduction of a new one to cause some eggs to he left unfertilized. 
November 18, 1897, I dug up live or six nests in comparatively still water where 
fishes had been seen spawning for a month. The sand and gravel were thrown into a 
screen and carefully sifted, but no eggs were found; but on stirring the gravel and 
cobblestones in a strong current and setting the screen below so as to catch floating 
objects, 13 eggs were secured; 11 were alive and all fertilized, 4 about 3 days old and 
the others about 28 days. I could see no indication of fertilization in the dead eggs. 
In November, 1898, in order to obtain eggs naturally spawned, I placed a screen 
obliquely in the Avater 2 or 3 feet below AA r here salmon were spawning. The screen 
was covered with small cobblestones, that the eggs might lodge among them and 
be protected from spawn-eating fishes. The first trial was unsuccessful. The second 
secured 48 eggs; 30 were dead when found. All of the live eggs were fertilized, the 
others could not be tested. 
The experiment with the screen is not a fair test for the number killed, as the 
screen caught much gravel Avhich pressed the eggs against the wires, and without 
doubt killed many more than would have been killed under natural conditions. The 
1897 experiment in securing eggs from the stream, Avhen 15 per cent \vere dead, Avas 
a fair test, but the number of eggs Avas too small to warrant definite conclusions. 
In both experiments all live eggs Avere fertilized. 
In November, 1900, 39 eggs were secured from natural spawning-beds in Battle 
Creek, 25 of them evidently fertilized, and 1 certainly not fertilized. The condition 
of the others could not be satisfactorily determined, as they were killed in securing. 
At another time a fish Avas artificially spawned in the creek on natural spawning- 
beds, a screen being placed beloAV to catch the eggs. A male Avas held in the same 
position immediately afterwards and milt expressed. The test Avas not quite fair, 
for although there Avas probably a larger quantity of milt than is discharged at once 
naturally, yet there was also a larger number of eggs. The eggs being caught by the 
screen and thereby remaining closer to Avhere the male was stripped was of no 
advantage, as the life of milt in water is ample to allow it to come to rest. If they 
had been farther away it would have given time for the milt to become more thoroughly 
disseminated through the water. The eddies caused by our standing in the water and 
holding the fish prevented the eggs and milt from floating off in a natural manner. 
In two trials, 35 per cent Avere fertilized in the first, and 50 per cent in the second. 
These various experimentsindicate a high percentage in natural fertilization, prob- 
ably over 80 per cent. It is significant that all live eggs found that had been spawned 
naturally, excepting one, Avere fertilized. The fertilization of dead eggs could not be 
determined, though they Avere no more liable to be unfertilized than live ones. 
Mortality among ova . — These experiments also point to a high but indefinite 
mortality from being covered by the gravel. The greatest loss, however, is probably 
due to spawn-eating fishes. In the. middle portion of the river, as at Battle Creek, 
when salmon are spaAvning, great numbers of other fishes, mostly the split-tail 
( Pogomclithys ), gather about to feed on the spawn. Fifty or more split-tails may 
often be seen lying a few feet downstream from a spawning salmon, and although 
each fish may eat but few eggs, all together they probably destroy a large per- 
centage of the eggs spawned in the middle portion of the river. 
Trout have been taken near the spawning platform at Battle Creek station Avith 
the stomach and throat gorged with eggs, the waste from artificial spawning, and 3 
