NATURAL HISTORY AND PHYSIOGRAPHY OF NEW BRUNSWICK. 339 
its course is here certainly pre- or post-glacial, though the former 
seems much the more probable. Its present course cannot, how- 
ever, be very ancient, since it runs directly across the prevailing 
northwest and southeast drainage characteristic of this region, 
though possibly it may run in a part of an old northeast-south- 
west valley as already noted. On the other hand there is evi- 
dence that a small part at least of its course falls in an ancient 
valley of the northwest-southeast series. An inspection of the 
map will show that the chain of McDougal Lakes, the Garrett 
Lakes, the small brook emptying into the Mitchell Lake outlet 
a portion of that outlet,* and one of the Caribou Lakes all lie in 
a line, which line, as the Crown Land plans show, points directly 
to the head of Indian Brook. I think it is possible we have here 
another of these ancient parallel northwest-southeast valleys, the 
third in the Walkemik system. 
Below the Mitchell Lake outlet the river flows through three 
or four deadwaters similar to those above, with the usual bould- 
ery rips between, then by a rough boulder-strewn stream, often 
with high rocky valley walls on one side or the other and with 
much drop, into the lowermost (and one of the largest) deadwaters 
on the river. Below this it becomes an extremely rough stream, 
one of the roughest in the province, falling heavily amongst huge 
boulders between steep banks of coarse glacial drift. It soon 
swings out of the prevailing southeasterly direction, and turns 
to the southwest, falling still more in a yet rougher channel, not 
only over boulders, but over schist and granite ledges crossing 
its entire bed. This part of the river below the bend is obviously 
post-glacial, flowing apparently in a trench 25 to 50 feet deep cut 
into level glacial deposits. Although I was not able to trace the 
pre-glacial course of this part of the Walkemik valley, I have no 
* My survey of this stream made it swing more to the east in its middle por- 
tion, indeed carrying it across the position of Lyle’s timber line ; but as that line 
shows no crossing streams, I have adjusted it to the position on the map. I had 
a somewhat similar experience in my survey of Mitchell Lake. I made it much 
longer than represented on the map, but as Lyle’s original plan of 1884 shows an 
apparently chained line, making the lake 1 1-4 miles in length along its axis from 
his north line, I have shortened my survey to fit these dimensions. It interests 
me to find, however, that a former letter of his to me gives the length of the lake 
as 1 3-4 miles. 
