suffered. Some coasts have a smooth seaward slope, and consist of imperfectly consolidated marine strata, dipping gently seaward, which have been little eroded since their emergence from the sea in which they were deposited: these are typical coasts of emergence. Other coasts, whatever their structure may be, exhibit forms of subaerial erosion, such as hills and valleys, the slopes of which appear to continue below sea-level, as if they had been partly submerged since they were eroded: these are coasts of submergence.

Coasts of Emergence. — Along coasts of emergence of the kind above specified the shore-line will generally be almost rectilinear or of simple curvature. The amount of emergence may be inferred from the altitude to which the marine strata rise along their inland border. It may be at once stated that coasts of this kind are seldom fronted by coral reefs, apparently because the loose sediments of their beaches and submarine slopes do not afford a suitable foundation for coral-growth: witness the Madras coast of India, the south coast of Java, and the west and south coasts of Borneo, all of which bear marks of sub-recent emergence. Another class of coasts of emergence, on which coral reefs abound, will be given special description below.

Young Volcanic Islands.—The coasts of young volcanic islands may be associated with coasts of emergence, especially if composed largely of ash and not of solid lava. They are frequently cliffed and beached, without reefs. Barren Island, east of the Andamans, in the Bay of Bengal, is somewhat cliffed, and but little fringed with corals. Réunion, in the western Indian Ocean, has reached a rather mature stage of erosion and abrasion, with a very imperfect development of fringing reefs, as will be further explained below. It therefore resembles certain strongly cliffed volcanic islands in temperate latitudes. Let it be noted that the cliffs of such islands are usually cut back by the waves at a faster rate than the valleys are cut down by their streams, so that the valleys are left hanging above sea-level, and their streams cascade down the cliffs to the beach.

Coasts of Submergence.—On coasts of submergence the shore-line will necessarily be irregular, advancing seaward around the outstanding points of partly submerged spurs and entering landward around the branching embayments of partly submerged valleys. Conversely, shore-lines of this kind indicate that the coasts which they border have been submerged, as Dana pointed out in 1849. Singularly enough, Darwin never perceived the value of this evidence in support of his theory (Davis, 1913).

The spur-ends of coasts of submergence in the coral seas usually offer excellent opportunity for the growth of fringing reefs, for their firm rocks are soon swept bare by the waves, and they are free from the detritus that accumulates in the bay-heads. If the submergence be slowly continued, a fringing reef, A (fig. 1), may be transformed into a barrier reef, B, by upward growth as the sea-level changes from S to T; but if the submergence be renewed at a more rapid rate, changing the sea-level from T to U, the barrier reef will be drowned, and, if a pause then occurs, a fringing reef of a new generation, G, will be formed, as will be more fully stated below.

Unconformable Reef Contacts.—In all cases of reefs bordering coasts of submergence the original fringing reef which forms the base of an upgrowing barrier reef, as well as the lagoon deposits within the barrier reef and the secondary fringing reefs that grow on the spur-ends of the lagoon shore, and also all fringing reefs of new generations, must rest unconformably on

an uneven foundation of subaerial erosion. This point has been too generally overlooked, although it is of the highest theoretical importance. Its converse is of practical value: reefs that rest unconformably on surfaces of subaerial erosion must have been initiated by submergence. Hence the nature of the contact of a reef and its foundation should be carefully observed, whether the reef be at sea-level or elevated above it.

Fig. 1.

Amount of Submergence.—The amount of submergence that an embayed coast has suffered is not well indicated by the depth of its embayments, for they may be much filled with sediments; the amount is better inferred by drawing a true-scale cross profile, as at P, fig. 2, of the spurs that enclose a bay-mouth, and continuing their slopes with decreasing declivity below sea-level until they meet. The visible cross-section of the valley above the

Fig. 2.

bay-head at Q should be taken as indicating the pattern of the submerged cross-section at the bay-mouth, P. The measure of submergence thus gained is only a minimum value, for, as shown in fig. 2, the depth of the submerged valley near the bay-mouth may be only about half the depth of the original valley-mouth, V.

Pre-submergence Period.—The duration of the pre-submergence period of subaerial erosion should be estimated as short, long, or very long, by comparing the actual form of the visible land-surface with its inferred initial form, due allowance being made for rock-resistance. In the case of dissected and embayed volcanic islands this comparison may often be made without much difficulty. On the coasts of continents and of continental islands the comparison may not be so easily instituted, but an attentive examination of the form of the coastal slopes will usually suffice to determine whether the cycle of erosion was in an early, middle, or late stage of its progress when it was interrupted by submergence.

Thus the submergence of the Queensland coast, in association with which the Great Barrier Reef of Australia and the discontinuous fringing reef in the broad lagoon were formed, did not occur until the rather resistant rocks which there prevail had been reduced to subdued forms of late maturity or even to the low relief of old age. The same may be said of much of the south-western coast of New Caledonia, except that the rocks there present near the shore are for the most part weaker than those of Queensland. In both these examples the pre-submergence period of sub-aerial erosion must have been of long duration.

In view of these various considerations it is evident that careful observation should be made of reef-bordered coasts from a physiographic as well as from a geological point of view, in order to determine whether the reefs have been formed in association with the submergence or the emergence of their foundation. It is also important that reef-free coasts in the coral seas should be similarly observed, in order to discover the conditions that do not favour reef-formation.

Rate of Submergence.—The ordinary statement of Darwin's theory of coral reefs implies that the rate at which reef-foundations have been submerged as a result of their own subsidence must not be greater, but may be less, than the rate of reef-upgrowth; and this has been held to be an improbable condition. Darwin's own statement of the problem made no such limitation as to the rate of subsidence, except where barrier reefs and atolls are actually found. For those reefs he stated that relatively rapid subsidences of small amount alternating with long stationary pauses probably represent the ordinary succession of events, and he believed that the average rate of submergence thus determined was not in such cases faster than the rate of reef-upgrowth.

This seems to hold true for the greater part of the open Pacific, where atolls and barrier reefs prevail, even though the submergence due to insular subsidence there has been accelerated by a sub-recent rise of ocean-level during the melting of the Pleistocene ice-sheets—a matter which has come into importance in recent years, as will be shown in more detail below. But exception to this statement is needed for an area to the north of the Fiji Group, where fifteen or more submarine banks, apparently submerged reefs or “drowned atolls,” have been discovered since Darwin's time; and also for the region of the Tonga Islands, where extensive submarine banks occur. In both these regions of the mid-Pacific, and in a few others, submergence appears to have taken place at a faster rate than reef-upgrowth. They thus correspond to a large part of the Indian Ocean, where submarine banks, apparently “drowned atolls,” prevail, as Darwin clearly understood.

Darwin on Fringing Reefs. — Furthermore, although Darwin regarded most fringing reefs as having been formed on stationary or on rising coasts, he clearly understood that rapid subsidence might drown earlier-formed reefs, whereupon the reefs that would grow on the new shore-line would be of the fringing class, as noted above. The statement of this point on page 124 of his Coral Reefs (1842) deserves attentive reading. True, inasmuch as Darwin did not understand that embayed shore-lines and unconformable reef contacts around spur-ends are sure signs of submergence, he discovered no examples of fringing reefs of this kind in the records that he studied, and all the fringing reefs on his chart are classed as occurring on stationary or rising coasts.

But his deductive expectation may now be confirmed, for the Australasian and other archipelagoes contain numerous examples of fringing reefs

unconformably contouring around the spur-ends of embayed coasts—witness Palawan, the south-westernmost member of the Philippines, and many other embayed islands in that group—well represented in recent charts of the United States Coast and Geodetic Survey; also the Andaman Islands, in the Bay of Bengal; for in all these examples the coast is elaborately embayed; and hence their fringing reefs must be unconformable, and their submergence must have taken place at a faster rate than reef-upgrowth. Many other examples of the same kind might be cited.

Fringing Reefs and Submarine Platforms.—Fringing reefs thus assume a much greater interest than is generally allowed to them: their relations to the features of the coasts they border deserve close attention. The breadth of the reefs should be noted as a means of estimating the time that has elapsed since the last movement of submergence took place. The off-shore soundings of reef-fringed coasts of submergence are also of importance, for they frequently reveal a submarine platform that in all probability represents a drowned barrier reef and its lagoon.

Such submarine platforms, several miles in width, are found in association with Palawan and the Andamans, although the sea-level fringing reefs of these islands are narrow. A well-developed submarine platform surrounds the greatly denuded “volcanic wreck” of Fauro, a small island with, narrow fringing reefs in the Solomon Group. A similar platform is shown by the latest surveys of the United States Hydrographic Office to surround the Samoan island of Tutuila; but the fact that the spur-ends of this island are rather strongly cliffed behind their fringing reefs distinguishes it from the other examples named. Submarine platforms occur around the Marquesas Islands also; but here, although the spur-ends are cliffed, as in Tutuila, they are not fronted by fringing reefs.

The depth of the submarine platforms off reef-fringed shores is not constant: along the west coast of Palawan the platform varies in depth from 25 or 30 fathoms near its southern end to 60 fathoms near its mid-length; the Fauro platform has depths of 70 or more fathoms; the Andaman platform is 30 or 40 fathoms in depth. On the other hand, part of the coast of Samar, in the Philippines, facing the open Pacific, has fringing reefs around its headlands, but its submarine slope descends rapidly to great depths. Now, let it be noted, first, that the three chief elements of the fringing-reef problem as here considered—duration of the pre-submergence period of subaerial erosion, rate and amount of submergence, and duration of post-submergence period of fringing-reef growth—have unlike values on different islands; secondly, that many other islands have well-developed barrier reefs which suggest slow submergence, and that some barrier reefs are broad and others are narrow, thus suggesting that the rate and date of their submergence are unlike; and, thirdly, that many elevated reefs occur at different altitudes and in different stages of erosion.

It thus becomes evident that the history of various reef-encircled islands must consist of unlike sequences of movements and pauses. Hence local movements of the reef - formations themselves, which may vary greatly, explain the varied facts much better than changes of ocean-level, which must everywhere be of the same rate, date, and amount. In order to learn how greatly the values of the various elements differ from place to place, their value for every coast should be determined independently. One of the most important of these elements is the

duration of the post-emergence or post-submergence stationary period, the estimation of which may now be considered in some detail.

Time since Emergence.—The shore-line of an emerged and thence-forward stationary coastal plain may be locally built forward, or “prograded,” by deltas if its rivers are of large volume and well charged with detritus from an elevated backland; and sand reefs enclosing shallow or marshy lagoons may be cast up by the waves between the deltas, and may advance seaward as the delta-fronts advance. Conditions of this sort appear to prevail along the Madras border of India, and around the south-west side of Borneo, thus proving that these coasts have been somewhat changed from their simpler initial form; but the littoral conditions are still manifestly unfavourable to coral-reef formation.

It is conceivable, however, that after a temporary supply of gravel and cobbles has been washed out by a flooded river to a certain part of the front of a delta that is for the most part composed of finer sediments the river may change its course, as rivers on deltas are prone to do. Then corals, attaching themselves to the larger cobbles, may spread sufficiently to form a small fringing reef, until a return of the river buries the corals. A buried reef of this kind will slant forward with the delta-front, and will lie conformably between the earlier and later foreest delta-beds. Such seems to have been the origin of a small elevated reef near Suva, Fiji: it lies on a local deposit of gravel, and both the gravel and the reef lie conformably in the slanting beds of volcanic mud, there known as “soapstone.”

The extent of the littoral lowland that is prograded along the border of a coastal plain will give some idea of the time that has elapsed since the plain emerged. But such lowlands are not always developed; for, if large rivers are wanting, the shore-line of a coastal plain may be cut back or retrograded farther and farther by the sea, as long as no change of level takes place. The farther it is cut back, the higher will be the resulting bluffs along the coastal-plain margin. The height of the bluffs along the shore of a retrograded coastal plain will therefore give an indication of the time during which it has been attacked by the sea. A more important point is that, however far such a stationary coast may be retrograded, a beach of loose detritus, continued off shore by a sheet of finer sediments, will, according to accepted physiographic theory, always cloak the abraded platform along the base of the retreating bluffs. No reefs are therefore to be expected on such a coast.

The Reef-free Coast of Madras.—It is important that the coasts of the coral seas should be examined with these principles in mind in order to test their correctness. As far as I have read, there is no published account of a strongly retrograded coast in the torrid seas that is still suffering abrasion in its original stand with respect to sea-level. It is interesting to note, however, that the high, hard-rock cliffs which, as described by Cushing, rise a short distance inland on the coast of Madras appear to have been cut back by the sea before the emergence of the present Madras coastal plain; hence the cliffs must, before the sub-recent movement of emergence by which a negative shift of the shore-line was caused, have exemplified a maturely retrograded, reef-free coast; and at the beginning of their abrasion the hard-rock land-mass must in all probability have been covered, near its shore-line at least, with the sediments of an ancient coastal plain of emergence, just as the emerged platform of marine abrasion which fronts the high cliffs is covered by a

modern coastal plain to-day; for otherwise it is difficult to understand why coral reefs should not have been formed there and have prevented the cutting of the high cliffs.

Cliffed Volcanic Islands.—It has been suggested above that the shore-lines of volcanic islands may be regarded as shore-lines of emergence, particularly if the island is largely composed of loosely compacted volcanic ash; for such a shore-line will be of comparatively simple outline, without pronounced salients or embayments, and the detritus washed down its slopes by its streams, added to that cut by the waves along the shore, will soon form a continuous beach, extending seaward in a sheet of loose sediments, on which reef-building corals cannot attach themselves (Davis, 1916B). Under such conditions the island will be continuously attacked by the waves, cliffs will be cut around its shore while valleys are eroded in its slopes, and if the island stand still long enough it will be completely truncated. Even then it may be difficult for corals to find a firm foundation for their growth until nearly all the loose detritus is swept off the surface of truncation.

According to Admiral Wharton, atolls were supposed to have been built up around the margin of truncated volcanic islands, no change of sea-level and no subsidence of the island being postulated. Darwin had previously considered this possibility and rejected it, because the resulting lagoons would be too shallow. According to Daly, atolls are supposed to have been built up on volcanic platforms that were abraded while the ocean was lowered and reef-building corals were killed, during the Glacial period. The best test of these suppositions involves a series of borings along the diameter of an atoll to a depth of 50 or more fathoms below present sea-level: the elevated atolls of the Loyalty Islands are to be recommended for such examination.

If it be true, as above suggested, that still-standing volcanic islands may, in the absence of protecting reefs, be cut away by the sea, a number of examples in different stages of abrasion should be found in the coral seas of to-day. Réunion is the best example of the kind, still in process of abrasion, that has come to my attention. Tahiti is an equally good example, but it has been somewhat submerged, and its shores are now defended by coral reefs, as will be more fully described below. Tutuila, in Samoa, and the Marquesas Islands probably, as noted above, belong to this series, but their place cannot be safely determined at present. Most volcanic islands in the coral seas are surrounded by barrier reefs, and their shore-lines are not cliffed. It is very desirable that all islands of the coral seas should be examined with the points here set forth in mind. The brief accounts now available of many such islands do not suffice to determine what stage of erosional and abrasional evolution they have reached.

Time since Submergence.—The headlands of coasts of submergence in temperate latitudes, not being defended by coral reefs, are vigorously attacked by storm waves; thus a cliff is formed rising high above sea-level, and a platform lying a little below sea-level. Coasts of submergence in the coral seas are as a rule fronted by barrier reefs or bordered by fringing reefs; hence they do not generally show the strongly cliffed spur-ends that characterize similar coasts in temperate latitudes. True, the spur-ends of such coasts are often cut off in low bluffs, B (fig. 4), 10 ft. to 50 ft. in height, forward from which one may see low-tide rock platforms 30 ft. to 100 ft. in breadth before one reaches the fringing reef, F, that is ordinarily found in such situations.

above sea-level. In the case of many volcanic islands it is not unreasonable to conclude that a barrier reef a mile from the island-shore has a vertical thickness of 1,000 ft. This conclusion evidently rejects the idea that a barrier reef is built upon a shallow platform of non-reef origin, which appears to me as improbable as that a volcanic island rests upon a shallow foundation of non-volcanic origin.

Mature Reef Plains.—Although coasts of recent submergence usually present favourable conditions for the growth of fringing or of barrier reefs, these conditions may not persist indefinitely on coasts that long remain stationary after a less recent submergence; for, if the land drained by the coastal rivers is of large-enough area, deltas, E (fig. 3), will in time not

Fig. 3.

only fill the drowned-valley embayments of a still-standing island, but will unite around the spur-ends and form a confluent alluvial lagoon plain, F, with a shore-line of comparatively simple pattern. As the advance or the progradation of such a plain continues, the fringing reefs on the spur-ends will be smothered with detritus. Such appears to have been the fate of many fringing reefs on the island of Tahiti, where an alluvial plain extends along much of the island-border. As the plain is still farther prograded, and as overwash of debris from the outer face to the inner slope of the outgrowing reef continues, the lagoon will be filled and converted into a mature reef plain, MP.

If the outwash of alluvium still goes on, the barrier reef, R, in spite of the width it may have then attained by outward growth, will be smothered, and its corals will be killed. Thereupon the sea will attack the reef and cut it away; and if this process be once begun there appears to be no reason for it to stop. The alluvial reef plain must in time be consumed, and then the central island will be attacked and cliffed; for as long as the island stands still, and as long as outwashed alluvium supplies material for a beach, coral growth cannot be re-established (Davis, 1917A). This sequence of events is evidently hypothetical in a high degree; nevertheless, the successive stages of such a sequence, and of all other reasonable sequences, should be carefully conceived by an observer of coral reefs while he is still on his voyage of investigation, in order that he may be able to confront the successive stages of the various sequences with the reefs that he sees, and thus discover which sequence gives the best history of their origin.

The danger of being buried and smothered in alluvium appears to threaten the barrier reef along the south coast of Viti Levu, Fiji, where the delta of the Rewa River has almost filled the lagoon; and a long stretch of the barrier reef on the south side of New Guinea appears to be

already smothered where the great delta of the Fly River has advanced far into the sea.

Most barrier reefs are, however, in no immediate danger of such a fate; their lagoons are far from being filled with alluvium; deltas, indeed, as a rule do not fill the embayments in which they head. Both the Rewa and the Fly are exceptional in being much larger than the rivers of most islands within barrier reefs. The prevalence of open lagoons shows that no such constant relation of land and sea level has been maintained as was provisionally postulated above, but that submergence has prevented lagoon-filling not only by providing new depths to be filled up, but also in the case of islands by diminishing the area and the altitude of the land from which part of the filling-material should come.

On the other hand, in view of the many possible changes of land and sea level, and of their many possible combinations with periods of rest, it is surprising that, among the many examples of reef-encircled islands, none are found with mature reef plains approaching or realizing the stage of smothering the corals on the reef-face. For if the development of barrier reefs depended only on the subsidence of their foundations, and if their foundations were of different ages and had subsided by different measures and at different dates, we should expect to see all stages of reef-development to-day—some close-set, discontinuous barriers; others broader and farther from their central island, the embayments of which should contain good-sized deltas; and so on through all the stages to a mature reef plain, the alluvium of which is just overlapping the reef; and then an old reef plain, much reduced from its original breadth by abrasion. But, with such exceptions as the Rewa and Fly deltas, no mature reef plains are known.

Combination of Island-subsidence with Changes of Ocean-level.—The absence of completed reef plains cannot be due to lack of detritus for their formation, for the amounts of detritus that have been discharged from many deeply denuded volcanic islands are vastly greater than the volumes of the lagoons enclosed by their barrier reefs. Hence the prevalence to-day of young barrier reefs with open lagoons must be taken as suggesting that some recent and widespread cause has produced a more general submergence than should be expected from island-subsidence alone; and this cause is perhaps to be found in the post-Glacial rise of ocean-level, for a rise of ocean-level combined with a prevalent but intermittent subsidence of reef-foundations would tend to maintain the barrier reefs of to-day in an early stage of their development and prevent the attainment of the more mature stage which they would reach during a long period of fixed levels of islands and ocean.

On the other hand, a fall of ocean-level, such as must have accompanied the oncoming of the last glacial epoch, would have tended to lessen or even to neutralize the submergence due to prevalent subsidence; hence during a glacial epoch lagoons may have been more generally filled than during an interglacial epoch or during the present post-Glacial epoch (Davis, 1915, p. 267; 1916A, p. 565). These somewhat transcendental aspects of the coral-reef problem are mentioned here in hopes that they may incite special observations, by means of which the possibilities here sketched may be assigned their proper values.

That the post-Glacial rise of ocean-level is not the entire or even the chief cause of the submergence under which barrier reefs as well as unconformable fringing reefs have been developed is proved by the great diversity

far as I can learn, low spur-end bluffs fronted by narrow rock platforms are of much more common occurrence than strong spur-end cliffs that plunge, except for their fringing reefs, into comparatively deep water. But fortunately certain islands possess strong spur-end cliffs fronting on deep lagoons: such islands deserve particular attention.

The Half-submerged Cliffs of Tahiti.—Tahiti, the largest island of the Society Group, is a volcanic doublet—that is, a larger and a smaller cone connected by an isthmus—now submaturely dissected by radial consequent valleys. In the central areas, where the valleys are close together and of great depth, the initial surface of the cones is lost; but it is well preserved in the peripheral areas, where the valleys are more widely separated and of moderate depth, as Dana long ago explained. The inter-valley spurs are, except at the north-western (or leeward) corner of the large cone, cut back in cliffs, which on the windward coasts rise 500 ft. to 1,000 ft. above present sea-level. Agassiz is, as far as I have read, the only observer of this beautiful island who has recognized the prevalence of cliffs around its shores.

Many of the smaller valleys are not cut down to present sea-level; their wet-weather streams fall in cascades from cliff-top notches. The larger valleys have been cut to a greater depth, for they descend below sea-level, and their mouths are occupied either by small arms of the sea or by delta-plains. The island is to-day bordered either by a fringing reef or by an alluvial plain, which is formed by the confluence of many delta-plains that have outgrown their valley-mouth embayments. Moreover, a somewhat discontinuous barrier reef now holds off the waves from most of the island circuit. Evidently, then, the cliffs have not been cut while the island has stood at its present level: they must have been cut when it stood relatively higher—that is, when the valleys were in process of deep erosion beneath present sea-level. Evidently, also, no reefs could have been present when the cliffs were cut.

The question then arises whether Tahiti stood still and had its cliffs cut while the ocean was lowered and chilled during the Glacial period, or whether Tahiti, besides experiencing changes of ocean-level in the Glacial period, itself subsided after cliffs had been cut around it, the cliffs having been formerly cut around Tahiti for the same reason that cliffs are now cut around Réunion—namely, because reef-forming corals cannot establish their colonies on the cobbles and gravels of the beaches that are ordinarily developed around the shore of a young volcanic island.

The latter alternative appears the more probable one of the two, for two reasons. First, the amount of the submergence by which the Tahitian valleys have been submerged appears to be 500 ft. or 600 ft. at least, and this is much more than the amount of lowering that the ocean is believed to have suffered during the Glacial period. Secondly, if the cliffs of Tahiti were cut around a still-standing island by the waves of the lowered and chilled ocean during the Glacial period, then the neighbouring island of Murea, as well as the other more distant members of the Society Group, should also be cut back in cliffs; but, apart from a few very exceptional cliffed spur-ends, that is not the case. The reefs of Tahiti should therefore be regarded not as having found their opportunity for upgrowth when the warming waters of the post-Glacial ocean were rising to their present level, but as having found their opportunity when submergence, caused in part at least by subsidence, embayed the island valleys so that the stream-washed detritus was pocketed in the embayments. In the absence of detritus the

its margin and if the larger part of its discharged detritus is to be deposited in the lagoon, G, of an upgrowing barrier reef, B; but in this case the early stages of subsidence must be so rapid, in order to provide sufficient lagoon-space for the deposition of detritus, that the upgrowth of a reef could hardly keep pace with it. It is not likely that the numerous barrier reefs of to-day have all survived so threatening a danger: hence a slower rate of early subsidence must be postulated.

Let the island, therefore, stand almost or quite still during a considerable period after its eruptive growth ceases. In this case the detritus supplied by the erosion of deep valleys, CY (fig. 5), and by the abrasion of high cliffs, CL, will be swept off shore in large amount, D, by vigorous waves, unimpeded by a barrier reef; then, if intermittent subsidence begin, placing sea-level at TE, the further discharge of detritus will be detained in the embayed valleys, E, and reef-upgrowth may begin. But, as under these conditions strong cliff-cutting will have accompanied the erosion of deep valleys, a considerable measure of subsidence, placing sea-level at UV, will be eventually necessary to submerge the cliff-tops, L, if they are not seen to-day. Whether this supposition represents the actual history of reef-encircled islands or not, it certainly provides a more reasonable condition for reef-growth than any other supposition here considered.

Fig. 6.

Various combinations of diverse conditions may be imagined. For example, the succession of events may be as follows: (1) Moderate cliff-cutting during a still-stand period before reefs are developed; (2) moderate submergence and reef-upgrowth; (3) a second still-stand period, resulting in the smothering of reefs by outwashed detritus, and renewal of cliff-cutting; (4) further subsidence and renewed reef-growth. Tahiti seems now to be approaching the third phase of this succession, for, if the present still-stand that is attested by the alluvial lowland around the island border endures as long as the earlier reefless period of valley and cliff-cutting, the lagoon will be overfilled, the smothered reefs will be abraded, and a new attack will be made by the waves on the cliffs at a higher level than before.

In any event, the only way of developing a barrier reef around a deeply dissected and non-cliffed volcanic island seems to be either to allow it to subside rapidly to a great depth while its reef is growing up, or to allow it to subside to a less depth after strong cliffs have been cut around its shore. And inasmuch as Réunion, Tutuila and the Marquesas, and Tahiti exemplify the second of these alternatives, the first alternative is regarded as the less probable of the two.

Many more observations of reef-encircled islands are needed before the questions here raised can be settled; and the observations must evidently be directed much more to the islands than to the reefs around them. The various possibilities here outlined, and as many others as can be invented,

stands near it, especially when the supposition of upheaval based on the occurrence of the volcano is contradicted by the occurrence of embayments in the shore-line of another island not far away.

The chief characteristics of crustal subsidences, as well as of crustal upheavals, are that the submergences or emergences they produce may vary from place to place in rate, amount, and date; and in these significant respects they will differ from the submergences or emergences due to other causes, which must involve universal changes of ocean-level, everywhere the same in date, amount, and rate, except where they are complicated by contemporaneous local crustal movements. Evidently, then, it is important to examine all structural and physiographic features of coral reefs and of their encircled islands from which inferences may be made as to the rate, amount, and date of the changes of level that they have suffered, in order to learn how far they are everywhere alike, or how far they vary from place to place.

Extravagant Deformation is demanded by Large Changes of Ocean-level.—A few examples of results already gained in this direction will be given below. But let it first be noted that in order to produce the submergence or upheaval of an island by 1,000 ft. a local subsidence or upheaval of the island by that amount in an ocean of essentially constant level is a much more economical movement than the vast crustal deformations involved in a rise or fall of the ocean-surface by the same amount around a still-standing island; for such a change of ocean-level can be brought about only by a change of the same measure in the entire ocean-floor (except around the still-standing island), or by a ten times greater change in a tenth of the ocean-floor.

Indeed, if a change of ocean-floor level over a tenth of its area involve roughly compensatory changes of a similar area elsewhere, then in order to cause a rise or fall of the ocean-surface by 1,000 ft. the failure of compensation must be of the order of 10,000 ft.; and, great as these movements are, their whole measure must be accomplished in the same period of time as that required for the much smaller measure of local upheaval or subsidence of the island under discussion. It thus appears that in seeking to account for a local submergence or emergence of 1,000 ft. an economy of vertical movements in a reef-encircled island involves an extravagance of movements elsewhere. Hence while small, slow, widespread, and synchronous changes in the relative level of land and sea may be plausibly ascribed to changes in the level of the ocean as a result of ocean-floor deformation, large, rapid, and local changes are best accounted for by movements of the island or coast where they are recorded.

Nevertheless, some students of coral reefs have attempted to throw the responsibility for large submergences or emergences of the islands that they have described upon other unspecified parts of the world. Thus C. W. Andrews says, regarding the emergence of Christmas Island, a little-dissected high-standing atoll, 1,200 ft. in altitude, in the eastern Indian Ocean, “It seems very probable that it is the general level of the surface of the sea that has been altered, and not merely a local upheaval of a limited land area that has taken place.” Inasmuch as in this case all islands and all continental shores that did not suffer emergence at the same time must have subsided with the ocean, an enormous terrestrial disturbance is involved in this method of accounting for the recently gamed altitude of a single small island.

difficult. It is of importance that the observer who has opportunity of examining a dissected reef should locate the structural details that he may discover with respect to the total reef-mass; it is also important that he should bear in mind the expectable structures of reefs formed according to the several chief theories of reef-origin, as shown in figs. 7 and 8, for he will thus be led to make special search for critical structures in their appropriate locations.

Fig. 7.

Thus if the great body of an elevated reef consist, as in fig. 7, of steeply sloping layers of reef detritus mostly free from admixture with volcanic sands and gravels, resting conformably upon a non-eroded volcanic slope, T, and more or less complicated by slides, the reef should be explained as a product of outgrowth during a prolonged still-stand period. Darwin clearly recognized the possibility of reef-formation in this manner, but regarded it as seldom occurring, because it would not result in the formation of a reef-enclosed lagoon from 20 to 40 fathoms in depth. Murray attempted to overcome this difficulty by assuming, as Semper had before him, that the lagoon-cavity would be excavated by solution; but the assumption is not supported by the features of lagoons, as has been noted above.

On the other hand, an elevated reef may show a three-part structure, as in fig. 8. The steeply dipping, exterior strata, T, may be formed of detritus chiefly derived from the reef, but with some fine sands and silts

Fig. 8.

from the central island. The slanting layers may be sometimes complicated by slide-structure as in the preceding case; they may rest on a heavy deposit of volcanic detritus, D, which should be associated with a buried cliff. The intermediate wall-like structure, R, should contain much coral in place, as well as large and small fragments. The outward or inward slant of the wall appears to be dependent on the rate of subsidence during

its formation (Davis, 1916C). The nearly horizontal interior strata, L, may contain coarse sand near the reef-wall, fine lagoon deposits in the middle, and volcanic sands and gravels near the central island; and the whole mass, with the exception of some of the outer slanting layers, may lie unconformably on a rock surface of subaerial erosion. In such a case reef-upgrowth during prolonged submergence probably due to subsidence would be inferred. Irregularities in the reef-wall, as in fig. 9, would indicate changes in the rate of submergence. A horizontal outgrowth, H,

would occur during a long still-stand period, when delta plains, E, might almost fill the lagoon. The occurrence of a buried cliff and platform in the profile of the underlying rock, and an exterior detrital deposit, as shown in figs. 2, 4, 5, 8, and 9, would be of much theoretical interest.

In case dissected atolls are found, their structures should be studied with especial care; and if their rock foundation is disclosed it should be closely examined to see whether the atoll limestones lie on it conformably or not. Christmas Island, in the eastern Indian Ocean, merits renewed study in this respect, for basalt has been seen in ravines behind its limestones, but the nature of the limestone-basalt contact has not been fully described.

Fig. 10.

In case elevated reefs occur in terrace-like arrangement, one above the other, as on Cebú in the Philippines, and elsewhere, the structure of the successive terraces will indicate the sequence of formation of their reefs.

The arrangement is as follows: The carotid and systemic arches are normal, but the pulmo-cutaneous arch bifurcates close to its origin, the posterior branch of which has the usual relations dividing into the pulmonary artery and the cutaneous artery. It is the anterior of the two branches which is exceptional, and appears to be the persistent third branchial arterial arch of the tadpole. This anterior branch arises from the base of the pulmo-cutaneous soon after it leaves the synangium: it is equal in diameter to this arch, and runs parallel with the other arches nearly to the point at which the last arch divides to form the pulmonary artery and the cutaneous artery. But as it approaches this cutaneous artery its diameter decreases and it bends backwards towards the fourth arch, to which it is joined by a very slender vessel. It then bends forward again and is continued into the cutaneous artery, alongside which runs as usual the petrohyoid muscle, to which it gives off twigs. The relation of this third arch to the cutaneous artery would suggest that the latter is derived from it, were it not for the precise account of the development of the latter given by Marshall. At the first bend of the arch is an angle as if a vessel or ligament passed forwards to the systemic arch, but I can find no trace of this. There is no connection between this third arch and either the systemic arch or the dorsal aorta.

On the left side I find that the condition of affairs is essentially the same, but the third arch is much more slender than on the right side. Less injection has penetrated the vessel, which suggests that possibly some resistance is exerted at the connection between it and the cutaneous artery. Nevertheless, the connecting vessel is distinctly red with injection, but is much narrower than its basal region. As on the right side, this third arch bends backwards (more abruptly than is shown in the drawing, for it is better seen when the arches are stretched apart) in order to reach the cutaneous artery, which on this side is normal and of equal diameter throughout its course.

The delicate pharyngeal artery, from the systemic arch, is plainly visible in the specimen below the third arch, but I have omitted it from the drawing for the sake of clearness.

According to Marshall,^* during metamorphosis “the third aortic arch in the third branchial arch of the tadpole atrophies altogether. In young frogs of the first year it loses its connection with the aorta and then gradually shortens up, the distal end becoming a solid cord, and the proximal or cardiac part retaining for a time its lumen. Before the end of the first year this vessel has entirely disappeared.”

In the larva this third afferent arch goes, of course, to a gill, and has no connection distally with the fourth arch: it is this union on the ventral surface that is rather puzzling in the present case, and especially the very slender union between the cutaneous artery and the parent fourth arch. It raises the question whether the ontogeny of the frog is really a true recapitulation of the phylogeny of the Anura, or whether the cutaneous artery is originally derived from the third arch, which in the embryology of those species of frog that have been studied has undergone some modification, leaving the third to become connected with the fourth arch at some stage in the history.

The cutaneous artery seems to be peculiar to the Anura, as no reference is made to such an artery in any description of the anatomy of salamander

Octochaetus thomasi Beddard.

Three individuals of this common South Island species were included in the collection from Stephen Island.

Maoridrilus tetragonurus Michaelsen.

This handsome species is evidently tolerably common on Stephen Island, as Michaelsen obtained four specimens. Ude speaks of several, and Dr. Thomson sent me seven individuals collected during his brief visit.

The largest specimen in this last gathering measures 210 mm. in length, which is not so long as those described by Ude, which attained as much as 280 mm.

I have nothing to add to the two accounts given by these two zoologists.

M. megacystis n. sp.

A single specimen of a small worm measures 90 mm. in length, with a diameter of 5 mm.; but it is poorly preserved, so that its dimensions are not accurately indicated by these figures.

Its colour is greyish-purple when preserved, and the clitellum has a redder tone.

The clitellum is fairly well marked over segments 14–22: that is to say, the segments themselves are glandular, but the intersegmental furrows still remain distinct.

The chaetae have the arrangement usual in the genus: the spaces aa, bc, and dd are practically equal, though aa is rather less than dd or bc—at any rate, behind the clitellum. Owing to the softness of the worm, it does not show the squareness of the tail which is to common in the genus.

Porophores are but feebly developed, and the ventral region of the segments 17 and 19 between the porophores is depressed so that in the 18th segment a slight transverse pad is left on the ventral surface. The spermatic grooves are convex mesially, and lie mediad of the ventral chaetae, which are quite distinct here, and are not thrust out of line of those in the neighbouring segments (as are those in the next species).

Internal Structure.—The septa separating the segments 8–14 are more or less thickened.

The dorsal vessel is single throughout the worm; the last heart is in the 13th segment.

The gizzard lies in the 6th. This is its true or “morphological” position, but, as is usually the case, it gets pushed backwards owing to the fact that the preceding region, like the gizzard itself, is longer than the segments to which it belongs.

There are large oesophageal glands in the 15th and 16th segments, and a smaller pair in the 14th. They are subspherical dilatations of the tube on each side, and the anterior two pairs meet their fellows above the gut.

The reproductive organs lie in the usual segments and in their normal positions.

Each spermatheca (fig. 1) has a relatively enormous diverticulum, which is as large as the ampulla—so large, indeed, that at first one thinks there are four spermathecae. The diverticulum, further, is not racemose, as usual, but has a smooth wall and a nearly globular form. When mounted, however, and viewed under the microscope one can see the outlines of the characteristic chamberlets into which its cavity is divided; but these walls do not affect the surface.

Another peculiar feature is that the duct of each of the two sacs opens into a large muscular “atrium,” which in its turn opens to the exterior.

The penial sacs are large, and the penial chaetae long. These are bluntly pointed, and the edges are curved upwards so as to form a short shallow furrow extending a short distance from the apex (fig. 2). These edges are ornamented with a few short bluntly pointed processes, but the rest of the chaeta is smooth. I examined not only a fully developed chaeta, but also one of the reserves, which exhibit precisely the same features, so that the processes are not produced by wear of the edges.

Fig. 1.— Maoridrilus megacystis. Spermatheca. a, muscular “atrium” common to the ducts of the ampulla and the diverticulum (d); s, septum.

Fig. 2.— M. megacystis. Tip of a penial chaeta, under a high power.

Fig. 3.— Maoridrilus thomsoni. View of the ventral surface of segments 15–21 (X 4 ½), showing the characteristic “trough” and the position of the ventral nephridiopores.

Fig. 4— M. thomsoni. Spermatheca.

Fig. 5.— M. thomsoni. Penial chaeta.

Locality.— Stephen Island.

Remarks. — The only species at present known in which the dorsal vessel is single throughout the body is M. michaelseni Ude (2), which was collected at Westport; but in that species the diverticulum of the spermatheca is quite small, its duct is narrow, and there is no “atrium”; further, the penial chaetae are quite different. The size of the diverticulum suggests the specific name.

M. thomsoni n. sp.

Of this species also there is but a single individual, which seems to be of about the same size as the previous species, but is in even a worse state of preservation than it. It is damaged just behind the clitellum, and is very soft. Its posterior end is missing, and there is as yet no sign of tapering. We do not know, therefore, what is its length. The fragment contains 198 segments and measures 65 mm. in length. The first 24 segments contribute 10 mm. to this.^* Its diameter behind the clitellum is

5 mm., but, as usual, is somewhat less at about the 7th segment, which is but 4 mm. across, and less again at the genital region, which is only 3 mm. It is evident that these measurements do not give a true idea of the dimensions.

The chaetae are equally spaced, so far as one can see on the animal.

The clitellum, though not yet thickened, seems to cover segments 16–20, for the intersegmental furrows are evanescent. Probably when the worm is mature the clitellum extends farther forward than this.

The spermathecal pores are conspicuous owing to their tumid lips.

There is one external feature in which this worm seems to show a marked peculiarity. On the ventral surface of the segments 17–19 there is a rather deep rectangular trough, with well-defined lateral and terminal boundaries, while the non-glandular floor is marked by longitudinal foldings. The appearance is that this ventral region is withdrawn by internal muscles (fig. 3).

The longitudinal margins correspond to the level of the ventral couples of chaetae, but on these segments, owing no doubt to the retraction of the ventral region, which results in the formation of the trough, the ventral chaetae and the nephridiopores are carried mediad of the line formed by these structures in the neighbouring segments. Under a dissecting-lens the chaetae themselves are not visible on these three segments, but the ventral nephridiopores of segments 17 and 19 are quite conspicuous and are out of the line.

The porophores lie within this lateral margin, and are not prominent. They project rather into the trough from the sides than from its floor, so that the pores face inwards towards the middle line.

The spermatic groove is very evident: its outer lip is formed on each side by the edge of the lateral wall of the troug; the inner lip is seen lower down this wall.

Internal Structure.—I did not note any specially thickened septa, as everything is so soft.

The dorsal vessel is single throughout the worm; the last heart is in the 13th segment.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

The gizzard belongs morphologically to the 6th, for the septum 5/6 is inserted at its hinder end; but its “apparent” position is in the 8th and 9th segments—that is, a transverse line across the body as dissected, passing over its anterior end, cuts through the intersegmental furrow and its posterior end lies at the transverse line through 9/10.

Oesophageal glands are large and spherical, and meet above the tubes in 14th and 15th, while in the 16th is a smaller gland; the intestine commences in segment 20.

The spermatheca has a large ampulla (larger actually than that of the previous species) with a narrow duct, which carries a small racemose diverticulum of the form usual in the genus (fig. 4).

The penial sac, and consequently the copulatory chaetae, are not nearly so long as usual. This, I think, is to be accounted for by the external trough, which probably aids in the process of copulation in this species, as it appears to do in certain other families of worms. But I can detect no “arcuate” muscles in these segments. The penial sac is not only relatively, but absolutely, smaller than in the previous species. Here it is scarcely longer than the length of the segment, whereas in M. megacystis it extends across the body-wall half-way towards the dorsal mid-line.

The penial chaeta (fig. 5) is slender, curved, and slightly swollen just below the apex, whose sharp point is slightly bent up to form a hook; it has no perceptible furrow. The sides are not ornamented by rows of minute teeth as in M. michaelseni, but in certain lights some five or six faint transverse lines can be made out just below the swollen region.

Locality.—D'Urville Island.

Remarks.—This species certainly has some resemblances to M. michaelseni, especially in the possession of a depression on the segments 17–19. But from it the present species differs in one or two features that seem to be specific. The gizzard, which in that species is said to occupy the two segments 6 and 7, here lies only in the 6th. I was careful to trace out the septa as above described.

The oesophageal glands, four in number, are said by Ude to be “small.” The penial chaetae are described as “long,” “spoon-shaped,” and ornamented with very fine teeth in transverse rows; and, though the tip is curved, its curvature is in the other direction, and there is no swelling below the apex. Thinking that perhaps this last feature was due to pinching with the forceps, I examined a “reserve,” or undeveloped, chaeta, which I find exhibits the same subterminal enlargement.

Ude also speaks of the penial sac as being “absent.” I have noted its very small size, and it may be that in a well-hardened specimen it would not project within the body-wall.

Had it not been, however, for the distinctness between the form of the penial chaetae in the two forms, I should have regarded this as merely a variety of Ude's species.

Perieodrilus durvilleanus n. sp.

A single individual was received, which unfortunately is immature.

A brick-red worm, with its mid-dorsal line of much deeper tone than elsewhere; each segment is marked by a number of white spots, in each of which is a chaeta.

Length, 108 mm.; diameter, 6 mm.; with 117 segments. The body is cylindrical, with scarcely any tapering at the hinder end.

The prostomium is tanylobic.

There are some 20–24 chaetae on each side of each segment. They are not in definite couples, but are more or less equidistant, though here and there a chaeta is absent.

The dorsal “gap” is about one-third the width of the ventral gap.

Dorsal pores are present, but I failed to note at what segment they commence. No nephridiopores are visible under the dissecting-lens, owing, I believe, to the softness of the wall.

There is no sign of a clitellum. On the 17th and 19th segments, outside the ventralmost chaeta on each side, is a faintly expressed papilla, recognizable in its immature condition by its pink colour in contrast with the nearly white colour of the surrounding skin. No spermatic groove is as yet present.

Internal Structure.—The septa behind the segments 9 to 13 or 14 are thicker than the rest

The dorsal vessel is double throughout the worm; enlarged hearts in 10th to 13th segments.

The gizzard is long, lying apparently in the 7th and 8th, but in reality it belongs to the 6th and possibly partly to the 7th.

The oesophageal pouches or glands are four pairs, in segments 10, 11, 12, and 13: each is a subglobular outgrowth marked by a series of vertical lines which indicate the attachment of internal folds or lamellae. They are quite lateral in position and do not overlap the gut.

The intestine commences in the 18th segment.

The gonads occupy the usual position on the hinder face of the septa of their respective segments, inserted close to the attachment of the septa to the body-wall. The prostates are as yet very small but quite distinct under a lens, and are of the usual form; the muscular ducts are recurved. There are no penial sacs, and at present no transverse muscles in these segments such as are present in the mature stage of the other species.

The spermathecae occupy the usual position: each consists of a pointed ovate sac, or “ampulla,” with a short stout duct, into the anterior face of which opens a bifid diverticulum, the free ends of which lie at the right

Fig. 6.—Perieodrilus durvilleanus. Spermatheca. The dotted circles on the diverticula indicate the chamberlets, which are visible only when the organ is cleared.

Fig. 7.—P. durvilleanus. The left side of segments 8 and 9, showing the spermathecae, nephridial tuft and its lateral extension, and in the latter segment the sperm-sac (s). The median line is towards the right side.

Fig. 8.—P. durvilleanus. The left half of segments 12 and 15, showing the gradual dissolution of the nephridial tuft and its extension dorsalwards.

and left sides of the ampulla (fig. 6). Externally they appear simple, but when mounted and viewed as transparent objects the internal chamberlets are visible. In this condition the diverticulum appears as a semicircular collar round the duct, much as I have figured it for P. ricardi (5); but seen in the animal the distinctness of the two long processes of the diverticulum is very evident.

There are four pairs of sperm-sacs, in segments 9, 10, 11, 12. The two anterior sacs have at present the form of long white slender cylindrical tubes resting against the hinder septa of the segments (fig. 7); each terminates upwards in a rounded end, and is attached ventrally to the septum close to the body-wall and nerve-cord. The two posterior sacs are shorter, wider, and lie along the anterior wall of their segments (fig. 8): in short, they have the usual position, but are at present only commencing to form.

The Nephridia.—From the 2nd to the 14th segments, inclusive, the nephridium is represented by a conspicuous tuft of minute looping tubules arranged in such a way as to form a sort of “rosette” close to the nerve-cord and occupying nearly the whole length of its segment. It is thus a more or less rounded or quadrate mass of tubules. These tufts are much larger in the more anterior segments, and in the 2nd, 3rd, and 4th seem to represent the entire organ, but farther back one sees that the tubules are not confined to these tufts but extend outwards along the body-wall for a short distance as a linear series of isolated loops near the septa (fig. 7). In about the 7th-11th I believed that under a high-power dissecting-lens I could detect a duct or tube passing outwards, and ending apparently on the body-wall about half-way up the side of the body. I therefore cut out, stained, and mounted a portion of the side of the body, including the segments 7–11, in the hope of being able to satisfy myself as to the locality of the pore; but I was unsuccessful. The body-wall is too soft to allow such a small aperture to be recognized.

I then mounted the cuticle of these segments, but was no more successful, for, though the large spermathecal pores and the linings of the chaetiferous follicles are perfectly evident, there is no pore that I could interpret as being the nephridiopore.

Sections were equally useless, owing, as I believe, to the soft condition of the specimen.

To continue the internal appearance: The dissolution of the tuft of the tubules, which commences about the 7th segment, continues till at the 15th almost all the loops are arranged in the linear series (fig. 8), and by the 17th I fail to see any tuft or rosette. At the 20th I am unable to detect any loops under a lens, but by picking up at random the tissue that lies between the septa I find under the microscope that it consists of minute nephridial tubules with accompanying blood-vessels.

I was unable to detect any funnel, but the poor state of the tissue has rendered it difficult to make as thorough an investigation on this important point as is necessary.

However, it is clearly, I think, a “meganephric” worm such as I have previously described.

Locality.—D'Urville Island.

Remarks.—The genus Perieodrilus (which Michaelsen has separated from my Plagiochaeta) (4) is so far confined to the mountains of the West Coast: it is therefore not surprising that a representative occurs in this island.

It is evident that the present species is nearly related to P. montanus and to P. ricardi (5), but from each it differs in one or more features. Externally its coloration recalls that of the former, as also in the concentration of the nephridial loops near the ventral region of the body (6); but in P. montanus the gonads are situated on the posterior wall of their segments, in P. ricardi they are on the ventral wall midway between the septa. Only in P. lateralis are they in their normal anterior position as in the present species, but in that worm there are no oesophageal glands and only two pairs of sperm-sacs. In the two other species, while there are four pairs of sperm-sacs, there is only one pair of oesophageal glands. The form of the spermathecal diverticulum likewise differs from that in the known species.

pass through a period of rapid growth followed by a period of comparative stagnation. This periodic growth has generally been attributed to changes of temperature and corresponding changes in the abundance of food-supply; and in regard to many species of fish it has been demonstrated that the maximum rate of growth roughly coincides with the maximum temperature of the water. There is evidence, however, to show that this periodic growth is well marked in the scales of some deep-sea fish, which can hardly be subject to any marked seasonal changes of temperature, and in the case of the squeteague (Cynoscion regalis) Taylor (6) has shown that the period of stagnation roughly coincides with the spawning season in midsummer. It seems probable, therefore, that the period of stagnation is determined more by a voluntary fast during the spawning season than by any actual shortage of food, and that individuals which have not arrived at sexual maturity subject themselves to this annual fast, though not to the same extent as the mature specimens. This voluntary-fast theory is further borne out by the observations of Masterman (5), who in a most careful critique of the previous work on salmon points out that a certain number of salmon captured at sea throughout the summer show no evidence of summer feeding. He concludes that some salmon start their spawning-fast many months before entering fresh water. This may cause the age of salmon to be underestimated in some cases, and certainly throws grave doubt on Johnston's claim that he can tell approximately the month of entering fresh water. In the case of trout there is no evidence of prolonged fasts, except during the spawning season, which occurs in midwinter, and it is of little importance whether the cause be lack of appetite or lack of food. There is some evidence to show that in Canterbury the maximum rate of growth, especially amongst the larger fish, occurs in spring rather than in summer. It is probably quite safe to assume, however, that the period of stagnation occurs in the winter.

Roughly speaking, a trout-scale (Plate I, fig. 1) consists of a transparent plate of more or less elliptical form, having its centre of growth approximately at one of the foci. Surrounding this and roughly concentric with the outer edge of the scale are a number of lines or “circuli.” The scale grows by the addition of these circuli round the periphery, which are added in greater numbers and more widely spaced during the periods of rapid growth. This alternate spacing and crowding produces light and dark zones, one light and one dark corresponding to a complete year's growth. The dark zones are called “annuli,” or “winter bands.”

In the case of spawning fish the stagnation is more complete, and the winter band is narrower and more clearly defined. In salmon (Salmo salar) the act of spawning leaves a clearly defined scar or “spawning-mark” on the scales, due to disintegration or reabsorption of the scale, especially along the lateral edges and the outer surface containing the circuli. A true spawning-mark is not very common in trout, but the character of the winter bands gives a fairly reliable indication of spawning. Plate I, fig. 1, shows one such winter band.

The exact cause of the spawning-mark in salmon is still in dispute. Johnston (4) attributed it to the vicissitudes of river life, whereby the fish shrank in girth, and says, “The compression of imbricated scales tends to increase the amount of overlap, and from this or dermic influences we find that their margins become ragged or frayed.” Masterman (5) has shown that this fraying or erosion in many cases starts long prior to the fish's entry into fresh water, and concludes that the phenomenon is one of

“erosion or absorption by the living tissue which is known to envelop the scale.” Two possible explanations are suggested: “The process may be an anticipatory reduction of the size of the scale to meet the approaching reduction in the girth of the body, or it may be connected directly with the formation and development of the ova.”

In Canterbury the spawning-mark is by no means so uncommon in trout as it appears to be in England and Norway. With male fish of considerable size (say, over 24 in.) it is rather the exception to find scales that do not show a definite spawning-mark—at any rate, in the Selwyn River (see Plate III, fig. 2). In females the act of spawning seems to leavea less decided scar, and most of the cases come within the region of uncertainty mentioned by Masterman, and introduce the personal element. In handling a large number of spawning fish this year, whilst collecting scales, I found that I could in almost every case detect the males by the texture of the skin. The males had a thick tough outer skin, and great difficulty was experienced in removing the scales, whilst no such covering was present in the females, and the scales were easily removed. Under the microscope the scales themselves were in many cases readily distinguished, those of the males being very much more eroded than those of the females. The ripe testes form a very much smaller proportion of the total weight of a male than the ripe ova of a female, so it is natural to suppose that the wastage of tissue in producing the former would be less than in producing the latter, and the shrinkage in milting is certainly less than in spawning, yet the scale-erosion is greater in males. All this seems to suggest that scale-erosion at spawning-time, in trout at any rate, is intimately connected with the production of the thick tough skin assumed by the males. Dahl has noticed that the erosion of scales in spawning salmon is more pronounced in the males, but apparently attaches no significance to his observation. In many cases it is a matter of opinion whether there is a spawning-mark corresponding to any particular winter on a trout-scale, but of the thirteen tagged fish from which I have scales every one shows, if not a distinct spawning-mark, at least a sharply defined winter band, such as the third winter band in Plate I, fig. 1, corresponding to the winter when the fish was stripped and tagged. I think it is probable that such winter bands are tolerably reliable evidence of spawning, but there is an almost perfect gradation from the broad ill-defined bands of the first two winters in Plate I, fig. 1, and many cases must always remain doubtful.

Dahl (1) assumed that the scales grew in the same proportion as the fish, and consequently that the distances from the centre of growth to the successive winter bands would be in the same ratio as the lengths attained by the fish in each successive winter. This assumption was almost in the nature of a corollary from what was previously known of the formation of winter bands, but experimental proof was desirable. Dahl and others have collected such a wealth of indirect evidence in favour of this hypothesis that there is little danger in accepting it as the basis of my investigations. Direct evidence, however, is difficult to obtain, and is meagre. As the whole of the present investigation depends on the truth of Dahl's hypothesis, it will be as well to add my small quota, more especially as Masterman and others have raised the objection that direct evidence is almost, if not entirely, lacking.

The North Canterbury Acclimatization Society annually strips a number of trout in the Selwyn River for piscicultural purposes, and takes the opportunity to tag two or three hundred fish each year with a small silver

label bearing a distinctive number; at the same time particulars of length, weight, and sex are recorded. Through the kindness of the society, and of anglers who have had the good fortune to recapture tagged fish, I have secured scales from thirteen of these fish when recaptured, and have calculated from these scales the length of each fish when tagged.

The following table shows the length when recaptured; the length (a) actual, measured at time of tagging, and (b) calculated from the scales; together with the difference in each case between the calculated and the measured lengths:—

In no case is the difference between the calculated and the measured length more than ½ in., and in only three cases is it so much, whilst in four cases the agreement is exact. Considering the difficulty of measuring two or three hundred live fish accurately, these results may be taken to fall well within the limits of experimental error in measuring. In practice the fish are generally measured to the nearest ½ in., and an error of ¼ in. at time of tagging and another ¼ in. when recaptured would be sufficient to account for the largest discrepancy of ½ in.

Two scales taken from one of these fish (tag No. 1374) at different times are shown (Plate II, figs. 1 and 2). The scale in Plate II, fig. 1, was taken on the 17th June, 1917, when the fish was tagged, and measured 21½ in. The scale in Plate II, fig. 2, was taken on the 28th October, 1917, when the fish was recaptured, and measured 23 in. The lengths each winter, calculated from a set of scales taken in June and a set of scales taken in October, are as follows:—

From the Acclimatization Society's records I have calculated the average length of the fish tagged each year since 1915. The figures in parentheses give the number of fish measured:—

1915—Males, 20.8 in. (100); females, 20.1 in. (98).

1916—Females, 19.3 in. (199).

1917—Females, 20.4 in. (140).

1918—Males, 22.6 in. (66); females, 21.5 in. (156).

In the years 1915 and 1918, when both sexes were tagged, the males averaged about 1 in. longer than the females. The average lengths of the samples from which I took scales are as follows:—

1915—Males, 21.3 in. (33).

1917—Females, 20.4 in. (140).

1918—Males, 22.5 in. (29); females, 21.7 in. (36).

These figures agree closely with the averages for the total fish measured, so the samples were in all probability fairly representative. The year 1918 was remarkable both for the number and large size of the spawning fish. The average ages [see Tables I (A) to I (F)] indicate that the males either have a shorter life, or cease to run up the river at an earlier age. This bears out the general belief that the spawning mortality is greater amongst the males.

Fig. 3.

A point to notice in these curves is that they are nearly straight lines for the first four years. This does not mean that each individual fish increases in length by approximately the same amount each year up to four years old. So far as my experience goes, growth of this character is almost unknown amongst trout in Canterbury, although such apparently is not the case in Norway. In Canterbury I have found that unless some outside influence is at work the rate of growth almost invariably starts to decrease quite appreciably in the third year, and this decrease is

The material which I have examined up to the present consists of a quite inadequate number of fish from several rivers of each class.

In Table II particulars are given of thirty-five fish from the Cam, North Branch of the Waimakariri, Styx, Selwyn No. 2, Opihi, and Tengawai, all of which I class as rain rivers. There are individual differences, but the average rate of growth in all these is very similar. The average for the whole thirty-five fish is—

In Table III particulars are given of nineteen fish from the Ashley, Waimakariri, and Rakaia. The figures for the Ashley and the Rakaia agree very closely, but the figures for the Waimakariri are nearer to those for the rain rivers. The probable reason for this is that four out of the seven fish were taken from the Belfast branch of the river, which is frequently very low, and probably contains only a small percentage of sea-going fish. The average for the whole nineteen fish is—

The averages for seven-, eight-, and nine-year-old fish are the figures for one fish only, an old jack from the Rakaia, which showed no sign of ever having been in the sea. The growth-curves plotted from these figures are shown together in fig. 4 for comparison—the rain-river fish by a continuous line, the snow-river fish by a broken line.

The points to notice are that, although the rate of growth is approximately the same in each class for the first two years, the rain-river fish fall off

rapidly in the third, fourth, and fifth years, whilst in the snow-river fish a good growth is maintained. The growth-curve for the snow-river fish is a typical curve for a sample of which the individuals have migrated to more favourable surroundings at varying ages. Probably no individual fish has a growth-curve of this type, but the average curve is compounded of several different types representing the one-, two-, three-, and four-year-old migrants respectively. The sudden jump during the sixth year in the rain-river curve is the result of two fish only, and no importance attaches to it.

It may seem rather arbitrary to include the Opihi as a rain river and the Ashley as a snow river. Each of these rivers is more or less on the border-line. Sea-going fish frequent each, and each contains a large number of small fish which have not been to sea, though possibly they may go later. It so happens that all my examples from the Opihi, which were caught in November, 1917, near the junction with the Tengawai, belonged to this latter class, whilst all the Ashley fish were largish fish which had probably been to sea.

In view of the exceptionally poor growth of the Opihi and Tengawai fish, it would be most interesting to get scales from some of the large sea-run fish for which the Opihi is so famous, and to see whether these represent a later stage in the development of fish which as three- or four-year-olds had averaged only 10 in. to 12 in., or whether they belong to a different race. The matter is of some importance to the South Canterbury Acclimatization Society. If these poorly developed three- and four-year-olds are practically the “parr” stage of the larger sea-going trout the present condition of the Opihi is healthy; if not, then in my opinion it is carrying a stock far in excess of its food-supply.^*

Plate IV, fig. 2, shows a scale from one of the Rakaia fish, and is an example of a very clearly marked scale, which is none the less difficult to read. The first winter band is clearly shown, and so is the second; but immediately outside the latter is another darkening; there is then a space indicating rapid summer growth, and the rest of the scale is normal. If this peculiarity existed in one scale only it might be attributed to some accident to or displacement of that particular scale. I have ten scales from this particular fish, and every one of them shows the same peculiarity. It has some meaning if one could only find it out. With some diffidence I offer the following explanation: The fish lived in the stream where hatched—probably the Rakaia—throughout the first year and the second summer and autumn; when the second winter band was nearly complete it migrated to the sea, and immediately responded to the stimulus of sea-water. The stimulus, however, was short-lived, and winter stagnation again set in, causing the third check. So the second and third checks are really one winter band divided by a short period of rapid growth in winter, due to the tonic effect of sea-water. I have met this same peculiarity in one or two other fish from the Rakaia.

Peculiarities such as this are not uncommon, and when they occur in one scale they invariably occur in every scale from the same fish, showing

one of the ten fish examined, except one, in which the growth was equal for the first and second years. Such a state of affairs would generally be explained by saying the fish migrated at one year old to more favourable conditions. This cannot be the explanation in Marymere, as there are no streams running either into or out of the lake. It must be remembered, however, that the insects in our back-country districts are mostly large in size, and it seems probable that the true explanation is to be found in the fact that the main bulk of the food-supply is of a nature more suitable to fish after they have passed the yearling stage. Gilbert (2) has shown that quinnat and sockeye salmon which have migrated to the sea as fry a few months after hatching have very similar scales.

No brown trout have been liberated in Marymere since 1908, and so it is clear that they must breed in the lake itself, as three of the ten fish appear to have been hatched in 1914. It is curious that the next-youngest fish seems to have been hatched in 1911. In dealing with such small samples it is dangerous to generalize, but it certainly looks as if 1914 was an exceptionally favourable breeding season. Plate V, fig. 1, shows a scale from one of these fish hatched in 1914, and Plate V, fig. 2, a scale from one of the older fish. The latter illustrates clearly the difficulty in determining the age of old fish, owing to the way in which the winter bands are crowded together towards the edge of the scale. It is probable that the percentage of ova hatched in Marymere, except in very favourable seasons, is abnormally low, and that the stock of fish is maintained mainly by the greater average age attained. Whether this latter is due to natural causes or to the limited amount of angling I am unable to say, but it is a characteristic not only of Marymere but also of other back-country lakes. The average age is 6.4, calculated to last winter; and as these fish would probably all have survived until next spawning season their average age would then have been 7.4 years, or about 1½ years older than the Selwyn fish.

Lake Heron.—The average growth is as follows:—

The full figures are set out in Table IV (B), and the growth-curve is shown in fig. 6 as a continuous line, the Marymere curve being reproduced

as a broken line for comparison. Although the size of the older fish approximates fairly closely to that in Marymere, the average length of the younger

fish is distinctly lower, especially at two, three, and four years old. It is also noticeable that the curve is nearly a straight line for the first four years, and that even in the fifth year the falling-off is not very pronounced. In dealing with the Selwyn fish I pointed out that a curve of this character was to be associated with a sample of fish containing individuals which probably migrated to more favourable conditions at varying ages. An examination of the figures in Table IV (B) shows the characteristic increase of growth to have taken place in every case, and examples can be found of one-, two-, three-, and four-year-old migrants. Lake Heron differs from Marymere in that there are several small tributary streams flowing in and one fair-sized stream flowing out of it. In the spawning season these are packed with spawning trout. No doubt also a large number of trout spawn in the lake itself. I have selected five fish from the fifteen which appeared to have scales similar in character to the Marymere fish, and calculated the average rate of growth as follows:—

Fig. 7 shows the growth-curve for these, and the broken line shows the growth-curve for the Marymere fish. There is a difference of 1 in. at three years and 0.8 in. at four years old; elsewhere they agree to within 0.5 in.

In dealing with such small samples the agreement is most remarkable, and suggests the probability that these fish were bred in the lake itself or migrated as fry.

The average age—6.93 last winter or 7.93 at next spawning—is again high—higher than Marymere.

Lake Alexandrina — This lake contains some very large fish, and the average size is certainly greater than in either of the last two lakes. Unfortunately, I have been able to obtain scales from only two fish; the figures for these are shown in Table IV (C).

Lake Coleridge.—This lake was first stocked with brown trout in 1868, and has for many years been noted for the exceptionally large size of its trout. I have been able to obtain scales from three fish only from this lake, but two of them are so remarkable that I have included a photograph of a scale from each fish. Plate VI, fig. 1, shows a scale from a fish of 10½ lb. captured about the 5th November, 1917. The length of the fish was not supplied to me, but would probably be about 27½ in., and I have made my calculations on this assumption. The fish seems to have been three years old in the winter of 1917, and shows a most remarkable growth since the last winter. The figures for each year are as follows:—

This scale apparently belongs to the Marymere type, and the fish was probably bred in the lake. It is considerably larger than any three-year-old I have ever heard of. The second fish was captured on the 10th March, 1918. A photograph of one of these scales is shown in Plate VI, fig. 2. The fish weighed 17 lb. and measured 34½ in. in length. The scales are, I think, the most beautifully marked and at the same time the most interesting in my collection. Surrounding the centre of growth are four winter bands close together, denoting four years of poor growth. These are followed by a year of growth which, so far as I know, is quite unique. There is another year of good growth, and then two years of moderate growth. The last winter band is right at the edge of the scale, and it is perhaps open to question whether this represents the winter of 1917 or the beginning of the 1918 winter. The difficulty of reading the scale is increased by the fact that every scale is more or less broken or worn at the edge. The fish was an egg-bound female, and in this abnormal state it is unlikely that she would grow much. On the whole, I think it more probable that the winter band right at the edge of the scale represents the winter 1917, and that there has been practically no growth since then (represented by one or two rings only), and that the fish was going back in condition when caught, as is evidenced by the frayed lateral edges of the scale. On this assumption the figures are as follows:—

The individual growth-curves for these two fish are shown in fig. 8. It is a fortunate coincidence that in three fish from this lake I should have

I take this opportunity of expressing my thanks to Dr. Chilton for much kindly instruction in microscopy, and to the North Canterbury Acclimatization Society and many anglers for assistance in collecting scales. The photographs are by Messrs. Leghorne and Colgan, of the Radia Studio, to whom I am much indebted for their infinite pains and trouble to secure the best possible results.

References TO Literature.

(1.) Knut Dahl, The Age and Growth of Salmon and Trout in Norway, as shown by their Scales (translated from the Norwegian by Ian Bailee), Salmon and Trout Association, London, 1910.

(2.) C. H. Gilbert, Age at Maturity of the Pacific Coast Salmon of the Genus Oncorhynchus, Bulletin U.S. Bureau of Fisheries, vol. 32 (Document No. 767), 1912.

(3.) C. Hoffbauer, Die Altersbestimmung des Karpfen an seiner Schuppe. Allgemeine Fischerei Zeitung, Jahrg. 23, pp. 341–43, 1898; Jahrg. 25, pp. 135-39, 150–56, 297, Munchen, 1900.

(4.) H. W. Johnston, The Scales of Tay Salmon as indicative of Age, Growth, and Spawning-habit, Fishery Board for Scotland, Ann. Rep., vol. 23, pt. 2, 1904; vol 25, pt. 2, 1906; vol. 26, pt. 2, 1907.

(5.) A. T. Masterman, Report on Investigations upon the Salmon, with Special Reference to Age-determination by Study of Scales, Board of Agriculture and Fisheries, Fishery Investigations, Series 1, Salmon and Fresh-water Fisheries, vol. 1, London, 1913.

(6.) H. F. Taylor, The Structure and Growth of the Scales of the Squeteague and the Pigfish as indicative of Life-history, Bulletin U.S. Bureau of Fisheries, vol. 34 (Document No. 823), 1914.

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]