92 REPORT—1843. 
the minimum assumes the character of a flat and gentle depression, coming 
on progressively earlier and earlier in the series (as at Flushing, Natchez and 
Washington). 
(H.) In all these but Natchez and Washington, the curve, after the cha- 
racteristic minimum, rises very rapidly, but at Natchez this rise does not take 
place. At Washington the series terminates at 7 p.M. on the 21st, so that it 
remains undecided what course the curve takes in the later hours. 
(B.) This sheet exhibits a well-marked and prominent group of similar 
curves, which occupies an extent of longitude from Flushing to Gardiner, 
Maine = 19 minutes. The depression of the curve at each station is one 
of the most interesting features. By inspecting the Table of Ranges it 
will be seen that the greatest oscillation was observed at Middletown, and 
it appears probable that the oscillations decreased on radii from this point. I 
have endeavoured to exhibit this by terminating each set of ranges with the 
range at Middletown. It is, however, a matter of regret that the observa- 
tions were not more numerous by which this interesting point might have 
been more strikingly illustrated. The descents as well as the ascents of the 
curves increase towards Middletown, with the exception of the curve at Gar- 
diner: this curve appears to partake of the character of the curve obtained 
at Montreal in its anterior portion. The curve at Montreal evidently does 
not belong to this system. 
(B.) The order of time in which the lowest altitudes of the barometer at 
the several stations were observed, is as under :— 
Plishing’ Ts ees LIBEL «ee SO POM. 
New Haven ........ 22nd .... 1 A.M. 
William’s College.... 22nd .... 2 A.M. 
Middletown ........ 22nd .... 5 a.m.? (series broken.) 
GeO a We eles p= 22nd .... 6 A.M. 
MSGINWIOD stare aie ayes 22nd ....10 A.M. 
(B.) The curves to the north of this group, namely, those obtained at 
Quebec and Montreal, are decidedly different, and that obtained at Natchez 
evidently belongs to a different system. The oscillation at this station being 
under ‘100, exhibits the diurnal oscillation. 
(H.) Lincline to regard both Natchez and Washington as belonging to 
this system, or at least on the verge of it; and if so, the minima there observed 
(though very feebly marked) ought to be added to the above Table, thus 
WateHezi wee ccs ste att a es OTe sl 
Washington ........ ais...) O Pei 
Sept. 1887.—Providence, Boston, Gardiner, Burlington, Flushing, New 
York, New Haven, Middletown, William’s College, Montreal—Sitka. Also 
a few hours of the end of the term at Quebec. 
(H.) Sitka is entirely disconnected and has already been sufficiently cha- 
racterized. All the other stations exhibit descending curves, and in all, the 
final rate of descent is much more rapid than the initial, so that they obviously 
belong to one system. 
(H.) The curves in this term are generally smooth, especially those for 
Providence, Boston, Flushing, New York and New Haven. Traces of the di- 
urnal oscillations prevail in most of them, though a good deal distorted in 
their epochs. The Montreal is the only curve which (abstraction made of 
these) offers much undulation. Jn this the descent is suspended during all 
the interval from 7 p.m. Sept. 21 to 2 a.m. Sept. 22, and replaced by a gentle 

