30 PROCEEDINGS: BOSTON SOCIETY NATURAL HISTORY. 
nants of seven terrace levels marking at least as many and probably 
more left-hand swings. The earliest record is given by the highest 
347-foot level (pi. 3, fig. 1). That plain, now limited in extent, 
skirts the left-hand rock hill — the limit of swinging. The once 
more extensive plain was nearly consumed by the river as it swung to 
the left later at the 304-foot level (pi. 3, fig. 2) carving a scarp in the 
higher 347-foot terrace. That scarp, the oldest to the left of the river, 
is now weathered with rounded edges and deeply incised transverse 
valleys, but thousands of years ago when the river flowed at its base, 
and the entire valley was filled to the 304-foot level, that high ter¬ 
race scarp was young, with a sharply defined straight edge. A third 
time as the river swung to the left it eroded the 304-foot plain leaving 
a scarp descending to the 291-foot flood plain (pi. 4, fig. 3). This 
third position of the river is the most hypothetical, and is based only 
upon the two limited scarps at the back of the triangular 291-foot 
plains (plate 1). The remnants of these early plains are assuming old 
forms. They distinctly represent the highest plains in the valley, and 
those that were first formed. 
Terrace 'pattern, middle stage .— Allow the river to continue its 
sweeping and swinging, and a terrace pattern of some complication 
may result. It is possible that the river may cross and recross its 
valley many times. On the northeastern side of the river (plate 1) are 
shown the results of at least eight swings including the present posi¬ 
tion of the West River; and this is probably not a full count. 
Having carved the three highest terraces, the river continued to 
degrade and to swing back and forth across its valley, carving and 
depositing four terraces at successively lower and lower levels as acci¬ 
dentally determined by rock barriers, short-cuts, or cut-offs. The 
complexity of terrace pattern of the middle stage is determined by the 
remnants of the three early and four later flood plains as represented 
in diagrams (pis. 3, 4, 5, 6). At a later return of the river to this side 
of the valley (pi. 4, fig. 4) it was flowing from 279 to 280 feet above the 
sea, and succeeded in almost completely removing the remnants of 
the 291-foot flood plain. The two small triangular areas shown on 
the fourth diagram (pi. 4, fig. 4) represent all that remains of the for¬ 
merly more extended 291-foot flood plain. So perfect are these tri¬ 
angular areas to-day that they seem almost like models. When the 
river returned for the fifth time to the northern side of its valley, it 
was flowing at the 262- or 263-foot level, and the scarp cut was 18 
