not prove that the salmon are guided by a marvelous geographical instinct which leads them to their parent river in spite of the fact that the river can not be found. The waters of Russian River soak through these sand-bars, and the salmon instinct, we think, leads them merely to search for fresh waters. This matter is much in need of further investigation; at present, however, we find no reason to believe that the salmon enter the Rogue River simply because they were spawned there, or that a salmon hatched in the Clackamas River is more likely, on that account, to return to the Clackamas than to go up the Cowlitz or the Des Chûtes.
Attempts have been made to settle this question by marking the fry. But this is a very difficult matter, indeed. Almost the only structure which can be safely mutilated is the adipose fin, and this is often nipped off by sticklebacks and other meddling fish. The following experiments have been tried, according to Mr. Davis:
In March, 1896, 5,000 king salmon fry were marked by cutting off the adipose fin, then set free in the Clackamas River. Nearly 400 of these marked fish are said to have been taken in the Columbia in 1898 and a few more in 1899. In addition a few were taken in 1898, 1899 and 1900 in the Sacramento River, but in much less numbers than in the Columbia. In the Columbia most were taken at the mouth of the river where most of the fishing was done, but a few were in the original stream, the Clackamas. It is stated that the fry thus set free in the Clackamas came from eggs obtained in the Sacramento—a matter which has, however, no bearing on the present case.
In the Kalama hatchery on the Columbia River, Washington, 2,000 fry of the quinnat or king salmon were marked in 1899 by a V-shaped notch in the caudal fin. Numerous fishes thus marked were taken in the lower Columbia in 1901 and 1902. A few were taken at the Kalama hatchery, but some also at the hatcheries on Wind River and Clackamas River. At the hatchery on Chehalis River six or seven were taken, the stream not being a tributary of the Columbia, but flowing into Shoalwater Bay. None were noticed in the Sacramento. The evidence shows that the most who are hatched in a large stream tend to return to it, and that in general, most salmon return to the parent region.
There is no evidence that a salmon hatched in one branch of a river tends to return there rather than to any other. Experiments of Messrs. Rutter and Spaulding in marking adult fish at Karluk would indicate that they roam rather widely about the island before spawning. A spawning fish set free in Karluk River was found three days later at Red River, sixty miles away on the opposite side of Kadiak Island.
The introduction of salmon into new streams may throw some light on this question. In 1897 and 1898, 3,000,000 young king salmon fry were set free in Papermill Creek near Olema, California. This is a small stream flowing into the head of Tomales Bay, and it had never previously had a run of salmon. In 1900, and especially in 1901, large quinnat salmon appeared in considerable numbers in this stream. One specimen weighing about sixteen pounds was sent to the present writer for identification. These fishes certainly returned to
