same as D and E, the differences being no greater than those which may arise from the action of a river of constant volume. The argument for difference of volume is as strong by the comparison of A and B with C as in the other two sections. It seems certain that the upper lakes did not drain through the St. Clair river during the making of the Erigan section. The great cataract has therefore had two independent periods of activity separated by the Erigan period. This much seems clear from an examination of the gorge alone.

The level of the water in the Ontario basin has undergone. several changes during the making of the gorge. At the beginning of the first Niagara period the water probably stood higher than now at Lewiston, but not long above the Iroquois level. Then it gradually fell away to a level 80 to 100 feet lower than to-day. All this was during the first Niagara period. The great cataract stopped either at Foster's Flat or at the upper side of the Whirlpool. There is still some doubt on this point, with probability, it now seems to me, in favor of the latter place.

From the Whirlpool up to a point a few rods above the cantilever bridge the gorge is narrow and shallow, and the latter character continues also below the Whirlpool to the lower end of Foster's Flat. It was presumably about at the middle of the Erigan period that the Iroquois beach was made at Lewiston, 135 feet above present lake level. The water at that time stood in the gorge at a level slightly higher than the water level at the foot of the present falls. This, with the smallness of the Erigan Fall, explains why this part of the gorge is shallow. After that the water fell away again to 80 or 100 feet below its present level during the early part of the present period of the cataract and then more recently was raised again to its present level by the Champlain uplift which elevated the outlet of Lake Ontario. These changes, with the harder quality and higher level of the Medina rocks in the bottom of the gorge, explain its salient peculiarities below the Whirlpool, and the changes in the volume of the river find. their correlatives in the history of the upper lakes.

The Succession of the Lake Stages.

In the paper on the second Lake Algonquin it was shown that the whole interval from the present day back to the time of Warren Gulf is completely filled by the series of events there described. The gorge of Niagara from the falls to the cantilever bridge suggests no change or variation in the volume of the river. But at and below the bridge th gorge grows narrow and shallow. Considering the uniformity of the geolog

ical structure, this feature is explicable only on the supposition that the great cataract was not then active, but was replaced by a much smaller one. That was the Erigan cataract.

In the lakes, the Nipissing beach, which is the latest abandoned shore line marking a critical stage, connects the St. Clair and Nipissing outlets, and was evidently made when both were flowing. But the Nipissing outlet is now 160 feet above Lake Huron, so that ever since that outlet was abandoned Niagara has had all the discharge of the upper lakes. The position of the Nipissing beach is such that the Chicago and Trent valley outlets are absolutely excluded. The Nipissing beach marks the activity of the St. Clair outlet and also that of the Nipissing outlet, but not of any other. In tracing back the order of changes we are therefore compelled to pass from the St. Clair to the Nipissing outlet, and so from the present lakes to the second Lake Algonquin. There is no escape from this step. This brings us to a time when Niagara did not receive the waters of the upper lakes,-the perfect correlative of the Erigan section of the Niagara gorge. But before the second Lake Algonquin there was Warren Gulf with its marine. straits. The Erigan Fall replaced Niagara during that time also. The simplest possible supposition as to the cause of these changes, and it is the only cause which is indicated by the facts, is, that there has been going on throughout the whole period a very gradual differential elevation, greater always at the north than at the south. This elevation covers all the changes which have taken place since the Chippewa beach was at sea level, and tallies, we may say, with half the work done in the post-glacial Niagara gorge. That part of the gorge below the Whirlpool is similar to the part above the bridge, and suggests the existence of a pre-Erigan great cataract, substantially like the present one, which again suggests a pre-Warren lake stage, substantially like the present.

And for this we have only to suppose an immediately postglacial condition of elevation at the north similar in a general way to the present land attitude and a progressive pre-Warren depression, to introduce in reverse order all the post-Warren changes. While that elevation existed the gorge was cut out from Lewiston to the whirlpool. The first marked change produced by the depression was the opening of the Nipissing outlet, at which the Niagara cataract began to weaken. Further progress of the depression closed the St. Clair outlet, and the erosion of the Erigan gorge began. The continuing depression next transformed the Nipissing outlet into a strait, and Warren Gulf was established. At the maximum of the depression the Chippewa marine beach was formed. The ero

sion of the Erigan gorge thus covers the duration of the first Lake Algonquin, Warren Gulf, and the second Lake Algonquin.*

If this is a true statement of the order and character of the changes, we should expect to find the strongest confirmation of it in the most recent beaches with confirmations less and less distinctly marked as we go backward in time. And so it is. In the article on the second Lake Algonquin are detailed the facts which seem to show the existence and boundaries of that lake, and that it filled the epoch of lake history immediately preceding the present. In the present paper have been stated facts in support of the view that the Chippewa beach marks the climax of a great marine invasion which formed Warren Gulf, and facts which show with less distinctness, but with some degree of force, the existence of the first Lake Algonquin. To this it may be added that if we accept the enlarged gorge below the Whirlpool as evidence of a pre-Erigan discharge of the whole lake system by the Niagara river, it will necessarily follow that between that stage and the Warren Gulf epoch there must have existed a first Lake Algonquin marking the transition from the first Niagara lakes to the Warren Gulf epoch, just as the second Lake Algonquin marked the transition from the Warren Gulf epoch back to the second Niagara system.

Chronological Conspectus.

At its maximum the great Laurentide glacier covered the whole area of the Great Lakes. By a correlation of the abandoned shore lines, moraines and outlets, and the gorges, recently submerged shores and rivers of this region the following order of events is made out for the post-glacial history of the Great

* An idea of the probable slowness of the recession of the Erigan Fall and the time required to make its gorge may be gathered from the following words of Mr. G. K. Gilbert (Letter in Nature for May 17, 1894). After mentioning the several accurate surveys that have been made of the Falls. he says: "It has thus become known, first, that in the middle of the Horseshoe Fall, where the principal body of water descends, the brink retrogrades at the rate of four or five feet per annum; second, that the American Fall, carrying a much thinner sheet of water, retrogrades so slowly that its rate is concealed by errors of survey." Speaking of the post-glacial period when the upper lakes had their outlet in some other direction and Niagara was fed only by Lake Erie, he says: During that epoch the volume of the river was so small that cañon-cutting was affected only by the feeble process now illustrated by the American Fall." This was the Erigan epoch. (See also Mr Gilbert's " History of Niagara River." Smithsonian Report, 1890) The brink of the American Fall is to-day almost exactly in line with the brink of the cañon wall above and below. While the Horseshoe Fall has receded from the center of the American Fall (about 3000 feet) the American Fall has receded at most not more than 100 feet and probably less than 50 feet, and it still strikes at its foot upon rock ledges and bowiders which it has not yet been able to remove. From the measured difference between the volume of the St. Clair and Niagara rivers it is found that the contribution of Lake Erie is about three-elevenths of the whole, and this, as Spencer states, is almost the same as the present volume of the American Fall. (See also L. Y. Schermerhorn, this Journal vol. xxxiii, April. 1887.) The true Erigan section of the gorge is probably about three-fourths of a mile long.

Lakes. They are set down in seven principal stages with transitions or critical stages between.

I. Glacial, ice-dammed lakes. Outlets at Fort Wayne, Chicago and other places. Beaches correlated with moraines in Ohio. Glacial lakes fall by stages as outlets change on withdrawal of the glacier-dams. Land relatively high in the north but slowly subsiding.

First Transition: By withdrawal of glacier the Niagara river is opened and the upper lakes become united.

II. First Niagara lakes. First epoch of Niagara Falls begins at Lewiston. For a short time glacial Lake Iroquois receives the water from Niagara. Shore lines of lower levels of this glacial lake washed over and obliterated by later marine invasion. Gradual depression of land continues at north, finally opening Nipissing outlet.

Second Transition: First two-outlet climax. Marked by the ALGONQUIN BEACH. (Possible subdivision here for supposed Trent river outlet.) Gradual northward depression continues. First epoch of Niagara Falls closes at the Whirlpool. Epoch of Erigan Fall begins.

III. First Lake Algonquin. Outlet eastward over Nipissing pass.

Third Transition: Gradual northward depression continues. Nipissing outlet brought down to sea level. Lakes become marine.

IV. Warren Gulf (rising stage). Marine waters fill the three upper lakes, the Ontario, St. Lawrence and Winnipeg basins.

Fourth Transition: Marine climax. Marked by the CHIPPEWA BEACH. Northward depression ceases and gradual elevation begins. Iroquois and Herman marine beaches made at the same time as the Chippewa. This was probably the climax of the post-glacial warm epoch.

V. Warren Gulf (falling stage). Gradual northward elevation. Irregular uplifts in the north deforming Chippewa and Algonquin beaches.

Fifth Transition: Nipissing outlet raised to sea level. Upper lakes become fresh.

VI. Second Lake Algonquin. Outlet eastward over Nipissing pass. Probably a small amount of local uplift at outlet in early stage.

Sixth Transition: Second two-outlet climax. Marked by the NIPISSING BEACH. Epoch of Erigan Fail closes at a point between 40 and 80 rods above the cantilever bridge. Second (present) epoch of Niagara Falls begins.

VII. Second Niagara lakes (present stage). Lake Superior becomes independent. Great Champlain uplift at the northeast. Formation of St. Clair delta begins and continues to the present time.

Fort Wayne, Ind., Feb'y 18, 1895.

ART. XXII.-Disturbances in the direction of the Plumb-line in the Hawaiian Islands; by E. D. PRESTON.

PROFESSOR ALEXANDER, Surveyor General of Hawaii has made a comparison of the astronomical and geodetic latitudes on the three principal islands of the group. In addition to this, since the observed latitudes at Kawaihae, Hilo, and Waiau (Mauna Kea) are consistent with one another the latter has been adopted as a standard and a comparison has been made between the observed latitudes and the Mauna Kea standard. This brings out some interesting deflections of the plumb-line. See Fig. 1.

There appears to be a disturbance of more than a minute in the direction of gravity at the south point of Hawaii (Ka Lae). At Kohala the plumb-line is deflected half a minute towards the south and at Kalaieha nearly as much towards the north, the disturbance being in both cases towards the mountain. The enormous deflection at Ka Lae (67") is also to the northward. This is evidently caused by the great mass of Mauna Loa,* which adds its effect to that of Mauna Kea and moreover is comparatively near to the astronomical station.

On Maui the same phenomenon appears. At Haiku there is a deflection towards the south and at Kaupo there is one to the north, and as before the astronomical latitude determined

*Both Mauna Kea and Mauna Loa are nearly 14,000 feet high.