to one level and one outlet which was south through the St. Clair and Detroit rivers to Lake Erie.

If the order of changes in the earlier glacial lakes was substantially as here outlined, it indicates that the front of the glacier retreated less to the eastward and more to the northward than supposed by Mr. Upham. And this agrees more closely with the terminal moraine series, with the distribution of bowlders and with the direction of the glacial scratches which characterize the lake region

The glacier advanced only to the outer terminal moraine, as marked upon the map. Its habit of flowing forward in the basins and lagging on the higher lands between is well shown by the series of terminal moraines made during the retreat. Excepting a few omissions of minor details and a few additions in the north, the moraines are put down as mapped by Professor T. C. Chamberlain. They show clearly how the lake basins shaped the ice front into lobes. Two of the moraines were laid down in glacial lakes, viz., those which pass through Findlay, Ohio, and Saginaw, Michigan. But neither of them shows any tendency toward a concave front line which is thought by some to characterize glaciers or glacial lobes which end in lakes or the sea.

The map of the earlier glacial lakes is based chiefly upon the works of Fairchild, Gilbert, Leverett, Spencer, and Lawson, to which is added a considerable amount of negative evidence gathered by the writer from the contiguous regions on the north.

Glacial Lakes Algonquin and Iroquois. The union of the three upper lakes and the fall of Lake Warren brought about a new order of things, although the lakes which remained were glacial. The water in the basin fell to the Rome outlet, and has been called Lake Iroquois by Professor Spencer, who first surveyed its old shore lines in Ontario. The lake in the three upper basins fell to a level below the lowest of Lake Warren and had its outlet through the St. Clair river at Port Huron. This lake has been called Lake Algonquin, and was so named by Professor Spencer, who first surveyed some of its abandoned beaches in southern Ontario. These two were the last of the glacial lake series within the area of the Great Lakes, and were probably held for a longer time at one level than any of their predecessors. With one exception (Glacial Lake Agassiz in the Northwest) Lake Algonquin was the largest glacial lake in North America, if not in the world. They are shown on a separate map merely for the sake of clearness and not because they were of a different age or origin. They were

simply the last and in many respects the most important of the glacial lakes, and it is only to avoid confusion that they are presented separately. Only one position of the glacial barrier is shown for each lake and in each case it is the last position which could have held up the lake. By a comparison of the two maps it will be seen that the interval between D D in the first map and B B in the second is a wide one and that the interval between F F in the first and A A in the second is still wider. Doubtless there were several intermediate halting places for the ice front in each case. But these intervals seem to imply a long life for both lakes, and that Lake Algonquin was considerably the longer lived of the two. Nothing in the glacial history of the region, so far as known, points to a contrary conclusion. A notable feature, especially of Lake Algonquin, is the smallness or narrowness of the ice dam that held it in its last stage. The country northeast of Georgian bay where the ice dam was situated is high and rough, and there are only two passes to the Ottawa valley which are lower than the Algonquin beach in that vicinity. These are close together, one-the deeper of the two-over Lake Nipissing and the other over Lake Tamagaming. The Ottawa valley itself is a deep narrow valley in a high plateau for many miles below the Nipissing pass. On visiting the ground it is easy to see that a comparatively small glacier or ice tongue creeping across the plateau from the north or descending the upper valley of the Ottawa might completely fill and choke up the narrow passage to the east. There are great moraines on the edge of the plateau south of the valley marking the presence of the ice when it was in just this position. The glacial scratches and the bowlders also show that the ice crossed the Nipissing pass from the north. Lower down in the Ottawa valley the ice in the later stages of its retreat flowed towards the southeast. From the direction of the ice motion it seems quite certain that the barrier B B of Lake Iroquois withdrew far enough to open the St. Lawrence valley some time before it left A A at the Nipissing pass.

For a considerable term in its earlier stages Lake Algonquin had its outlet southward through the St. Clair and Detroit rivers to Lake Erie, and thence through the Niagara river to Lake Iroquois. But when the ice front had drawn back as far as Balsam Lake in Ontario the outlet shifted to this point and the overflow of Lake Algonquin passed down the Trent valley, to Lake Iroquois, or possibly to Lake Ontario after Lake Iroquois had been drained off. The exact relation of these two outlets and their comparative duration in time have not been worked out. It seems probable that

they were very nearly at the same level so that the change from one to the other did not make much difference in the level of the lake. As we shall see later, however, there is some reason to believe that there was an uplift of the land at the northeast closing the Trent valley outlet and throwing the discharge back again to the St. Clair river, and that this was the last outlet and was active during a considerable time preceding the withdrawal of the barrier from the northeast.

Perhaps the most remarkable fact about the old shore lines of the Great Lake region is the fact that they are not horizontal when compared with present water levels The Lake Warren beaches at the western end of the Erie basin and the Algonquin beach in the east half of the Superior basin are substantially horizontal. But all the rest are more or less inclined upward in a northeasterly direction. The inclination is not the same in different beaches, being generally greatest in the older and higher ones; and it varies considerably in the same beach in different places. That all the beaches were horizontal when they were made seems certain. It follows that their present departure from that attitude is the measure of the upheavals of the land since they were made. The older, higher beaches record the net result of many changes. But the lower, younger beaches record only such changes as occurred after they were made. Hence the deformation of the latter is generally simpler and shows fewer irregularities.

The altitude of the Algonquin and Iroquois beaches above the sea are shown by small figures on the map at a number of points where they have been measured. The figures in parenthesis show the altitude of the present lakes above the sea. The Algonquin beach passes southward under the present level of Lake Michigan, and if it continues at a uniform rate it strikes about 110 feet below the lake at Chicago. It seems certain, therefore, that Lake Algonquin never had its outlet at that place. This beach passes under Lake Huron also in Saginaw Bay and at the extreme southern tip near Port Huron. Prof. Spencer calculates it to be about twenty feet below the lake at the latter place, but it may be less, for the declivity of the beach decreases southward from Georgian Bay along the east shore. The slight submergence of the beach at Port Huron does not prevent the St. Clair and Detroit rivers from being an outlet, provided the declivity of the old Algonquin water level continues decreasing southward.

This outlet and Lake Erie approximately as it was then are shown on the map in heavy broken lines. Lake Erie was much smaller than now.

ment. Flowers and fruits should not be begun as separate lines until work on the plant element shows that they have been made of leaves. Stem structure and leaf structure form somewhat independent lines of work; the former may be taught❘ in the fourth grade, the latter may be begun in the fifth grade. As to plant physiology, plant distribution, and useful products the mind catches an enlightening hint now and then from all other lines of work.

It may be necessary to indicate the relations between the above generalizations and lines of work. A study of the plant element leads to a knowledge of the embryo and the morphology of the leaf and of the stem. Leaf structure, stem structure, roots, flowers and embryo lead to the correlation of structures in exogens and endogens. That leaves make plant substance is a generalization of plant physiology; the first hint of this is in germination, where it is seen that the plantlet needs food until it puts forth a green leaf. That flowers make seeds, grows out of flower study. The study of fruits shows how plants provide for the dissemination of seeds.

First line of work.-The plant element.

The following inductions are arranged in a graded series. Each one represents a certain complement of work done by the pupils and directed by the teacher.

1. A branch, with its leaves, consists of a succession of plant elements.

2. (In the first examples.) The plant element consists of internode, node and leaf.

3. Some plant elements consist of internode, node and a pair of leaves.

4. A plantlet consists of root, stem and leaves. 5. The ascending part of the plantlet is made up of a succession of plant elements.

6. Roots always grow from lower end of the first internode, (caulicle).

7. (As shown in germination.) The embryo consists of one or more plant elements.

8. The plumule is a little terminal bud and is made up of one or more plant elements.

9. (Omitting exceptions.) Buds grow out at the nodes and are either terminal or auxiliary. 10.

Bud scales are transformed leaves, and a winter bud consists of a succession of plant elements. (Best shown by buds opening in the spring.)

11. (As shown by branches of bedstraw, Galinus.) In some cases the plant element consists of internode, node and a whorl of leaves.

12. The turnip (during its first year) or the common thistle (during its rosette-leaved winter stage) consists of root, stem and leaves, but the stem is very short and the nodes are crowded.

23. At all nodes we find leaves; at some, leaves and buds; at some, leaves and roots; at some, leaves and buds and roots.

24. The plant consists of root, stem and leaves, and nothing more.

Only the prominent inductions have been mentioned, enough to show the order of movement. The first one is to the pupils but little more than an individual notion, and this is as it should be. Each pupil should have a switch with the leaves on, and all the switches should be of the same species. There is always danger of confusion from variety of examples. Study one thing at a time and compare it with related things previously studied. This first notion of the plant element in some particular kind of switch has content rather than extent; but new examples come up day after day, month after month, year after year, until the content all but disappears and the extent becomes very great. Allow the extent of a notion to grow; never declare it, at any stage. The first notion arises from an attempt to divide the branch into similar parts, (nature study). The second notion arises from an attempt to divide the plant element into parts. That it consists of internode, node and a single leaf implies that it is the simplest form of the plant element, and that the first examples should be branches bearing alternate leaves. No common terms can take the place of internode and node (the term joint is ambiguous), hence we are obliged to use these strange names as soon as the ideas that they represent are developed. The statement that some plant elements consist each of internode, node and a pair of leaves implies that a set of opposite-leaved branches have been worked. That the embryo consists of one or more plant elements is easily worked out by observing the germination of seeds. It shows that the plant be