But on the whole, we felt that the results of our European expedition were incomplete, because nothing like a full account of the gross 
Fig. 5. Restoration of Tremataspis, an Ostracoderm from the upper Silurian rocks of the Island of Oesel, Baltic Sea. anatomy of any one species of ostracoderms was obtained. The organs about the mouth and the location of the principal viscera and of the nervous system were entirely unknown; and it was quite essential to obtain evidence on these points, because it had become increasingly clear that this strange class of animals could not be safely interpreted in terms of either vertebrate or invertebrate anatomy.
On my return to America it was decided to make an attack on a younger branch of the ostracoderms; one that was known to occur in the Devonian rocks of the Bay of Chaleur in Canada.
Four summers were spent in this locality, in search of specimens well enough preserved to be used for anatomical study. Fragments, or in some cases nearly the whole head, could be readily found on the beach at low tide, or by splitting open the disc-shaped nodules that had been washed from the adjacent cliff. But these specimens were generally crushed out of shape, or were so badly weather beaten and worn that they were of little value. We hoped to find in the cliff, which extended along the water front for several miles, unweathered specimens that would show not only the whole head, but the rest of the body, the nature of which at that time was entirely unknown. We accordingly examined with great care the face of the cliff as far as it was accessible; and many tons of rock were dug out of it and split open in order to locate the particular beds that contained the fossils; but the latter appeared to be very irregularly distributed, for we did not succeed in finding a single specimen in that way.
At last we found, close to the foot of the cliffs, a large piece of rock that contained several fossils of the kind we were looking for. It could not have been carried there either by the waves or by drifting ice, for evidently it had fallen quite recently from the rocks above. It did not take long to locate, about twenty-five feet above this fragment, the beds
