the shingle. If this were a correct supposition, it was considered possible to locate the outcrop of the water-bearing stratum by means of observing the rainfall at a number of places between Lincoln and the mountains, and noting at which places the rainfall most nearly corresponded with the fluctuations in the level of the well. For this purpose rain-gauges were installed or existing installations were used to obtain records of the daily rainfall from the following places: Rolleston, Lawford, Kirwee, Darfield, Hororata, Glenroy, and Mount Torlesse. The positions of these places are shown on the following map, which also shows the two rivers of the district. The slope of the plain is from north-west to south-east.

Fig. 3.—Diagram of the Structure of the Plain.

Fig. 4.—Map of Portion of Canterbury Plain, Showing Positions of Raingauges.1, Lincoln; 2, Rolleston; 3, Lawford (half-way between Weedon's and West Melton); 4, Kirwee; 5, Darfield; 6, Hororata; 7, Glenroy; 8, Mount Torlesse (two miles above Springfield).

Unfortunately, all the records did not begin at the beginning of the year, those from Darfield not commencing till the 1st April, and those at Rolleston till the 1st June. Valuable information was thereby lost. The records are, however, complete and accurate for the periods they cover. In the accompanying graph (fig. 5) the averages of all the readings of the well for each week for five months are shown, and underneath them, the total weekly rainfalls for each of seven stations, Rolleston readings not having been commenced. Zero for rainfall is made a sloping line, roughly corresponding to the graph of the well, for the purpose of bringing the rainfalls - graph near to that of the well, to facilitate comparison. To simplify the figure, rainfalls are shown for only those weeks from which conclusions may be drawn. The falls of Glenroy and Kirwee were identical for the weeks shown, and therefore these two stations are represented by only a single symbol—viz., do [ unclear: ] s and dashes.

Starting with the station nearest the mountains—viz., Springfield—if the rainfall at Springfield (squares) for the week ending the 15th January was responsible for the rise of the well shown for the week ending the 22nd January, then also the much heavier rainfall for the week ending the 26th February must have been responsible for the almost imperceptible rise for the week ending the 5th March. These two results are inconsistent, and therefore it may be stated that the rainfall on which the well depends does not fall at Springfield, nor does the water-bearing stratum tapped by the well outcrop there. Similar inconsistencies may be noted for other localities, as follows:—Hororata: In the week ending the 15th January a rainfall of 1 in. is followed by a rise in the well of 1.5 cm., and on the 23rd April; a rainfall of 2½in. is followed by a decline of 1 cm. Glenroy shows inconsistencies for the weeks ending the 15th January and the 23rd April; Darfield for the weeks ending the 2nd and the 23rd April; Kirwee for the weeks ending the 15th January and the 23rd April; and Lawford for the weeks ending the 15th January and the 23rd April. But when we come to examine the rainfall at Lincoln, and compare that with the subsequent rises, or arrests of the decline of the graph of the well, a remarkable degree of consistency is disclosed. The graph of the static level of the well is as nearly parallel to that of the Lincoln rainfall as could possibly be expected under the circumstances, and, being given the rise of the well due to the rainfall of the 15th January, the graph of the one could be constructed with reasonable accuracy from that of the other. From this it is evident that the stratum tapped by the well outcrops in a district with a rainfall during the months shown almost exactly equal to that of Lincoln. None of the stations recording for me shows such an equality, and so it seems evident that the water-bearing stratum under consideration out-crops nearer to Lincoln than to the nearest of the stations. That station is Lawford, nine miles away, and so one would probably be safe in saying that the stratum of shingle 341 ft. under the surface at Lincoln reaches the surface seven miles or less up the plains. This would place the outcrop somewhere about Rolleston—a district noted for its loose shingly soil, directly underlaid by coarse gravels, with no interposing layer of clay. Such country is absorptive of water in the highest degree, and an ideal catching-area for an underground water-supply.

The surface of the land at Rolleston is 134 ft. above that at Lincoln. The water-bearing stratum then rises 475 ft. in seven miles, or about 68 ft. to the mile. The surface of the plains near their upper limit has a fall of about 60 ft. to the mile, while between Rolleston and Lincoln it is only

20 ft. to the mile. At the time that the fall on the surface of the plains between Rolleston and Lincoln was 68 ft. to the mile—that is, when our water-bearing stratum, was deposited—the whole plain must have had a much steeper gradient than at present. This would probably be due to the much greater supply of waste to the above-gorge waters of the rivers, so that in those times the present plains would have been much more like the present-day shingle fans than like plains. That the gradient of the plains was once much steeper than now is proved by the high terraces round Woodstock, and by Racecourse Hill, a residual shingle mound some 60 ft. high. The cutting into their beds of the present rivers is merely a continuation of the process of lessening the gradient of the plain, the bed of the Waimakariri being virtually level with the plains at their lower edge, and over 300 ft. below them at their upper limit. It is therefore in accord with what I suppose would be the expectations of geologists that at one time the surface of the plains should be much more steeply inclined than now, but that the supply of waste should be so great as to form a deposit sloping nearly 70 ft. to the mile forty miles away from the gorge is perhaps noteworthy.^*

It was stated above as evident that the collecting-ground for the well is nearer Lincoln than the nearest rain fall-station is. On the part of one unacquainted with the country, a possible objection to this is that the collecting-ground might equally well be more distant from Lincoln than the farthest station is. The country between Springfield and the West Coast, however, consists of mountains of greywackes and slates quite impervious to water in large quantities, and, in any case, this water would percolate out into the rivers flowing at the base of the mountains. The amount of water in the Waimakariri is, moreover, a gauge of the amount of rain falling on these mountains, and I have been so fortunate as to be supplied with daily readings of the height of the river during several months. Most of the floods that my records show occurred nearly contemporaneously with considerable rainfalls on the plains, and the subsequent rises of the well could not, therefore, be stated as dependent on, or independent of, the rises in the river. On the 23rd March, however (see the arrow-head in fig. 5), there was a heavy flood, sufficient to stop the mails at the Bealey, but, as fig. 5 shows, there was no sign of any rise or arrest of the decline of the well until rain fell in the second week after the flood.

Although it is impossible that the rain falling on the mountains should directly find its way into the water-bearing stratum tapped by the well, it seemed quite possible that after reaching the river the water might percolate into such a stratum where the river runs across its outcrop. This, indeed, is probably the common opinion held; but the observations made do not support the supposition, as far as the well at Lincoln goes. The observations on

The graph of the well is shown by the full line, and its variations in centimetres; that of the barometer by a dotted line, and its variations in inches. It will be observed that the scale for the well increases upwards, and that for the barometer downwards; further, the space between 29–82 in. and 30.30 in. on the barometer scale (virtually ½ in.) is the same as the space between 86 and 82 on the well scale—that is, 4 centimetres (virtually 2 in.). This means that the close agreement of the two graphs seen in fig. 6 has been obtained by turning the barometer-readings upside down and multiplying by four, as above stated. Figure 6 is a portion of a graph that was constructed for the whole year for the purpose of establishing the agreement between the fluctuations of the well and the barometer.

The section 22nd March to 8th April was chosen for illustration because it is fairly typical of the whole graph, and because there is no complication due to rainfall. There were showers on the 29th, 30th, and 31st March, and on the 1st, 3rd, 4th, and 6th April, but the three heaviest of these were only 0.07 in., 0.08 in., and 0.09 in., and the remaining were 0.01 in. each. These numerous rains, the greatest of which did not reach 0.1 in., cannot be suspected of influencing the graph of the well, since in any case the well does not always rise after the rainfalls shown.

That a low barometer is accompanied by a rise in the static level of shallow wells has been excellently demonstrated by F. H. King, and for artesians is recorded in the following sentence occurring in an article by Professor J. W. Gregory in the “Journal of the Royal Geographical Society” for August, 1911, p. 171: “The Hon. E. W. Lamb” kindly tells me that an increased flow has been observed in some of the wells of New South Wales at times of low barometric pressure. The increased flow from springs when the barometer is low is a well-known phenomenon which has been established, for example, by the work of Mr. Baldwin Latham near Croydon. The increase is no doubt due to gas-pressure, the gases dissolved in the water expanding when the atmospheric pressure is reduced. Mr. Latham's evidence therefore shows that gas-pressure acts even on wells of which the flow is mainly determined by ordinary water-pressure.”

I have examined this theory of the rise of wells under diminished atmospheric pressure. It appears to assume that water is compressible, or, at least, that the gases within the water are compressible. This, of course, is not so. If the pressure is diminished the gases will remain dissolved if the water is not already saturated with them, and if the water is saturated the gases will come out of solution and form bubbles. The water in the College well is saturated, contacting 26.50 c.c. of gases per litre at N.T.P., made up as follows: Carbon-dioxide, 1.07 c.c. per litre; oxygen, 4.29 c.c. per litre; nitrogen, 21.14 c.c. per litre.

It must be further remembered that the volume of a gas absorbed by water is independent of the pressure, since, although doubling the pressure doubles the mass of the gas absorbed, the same doubling of the pressure halves its volume. If, then, the pressure were suddenly diminished the volume of gases liberated would be proportional to the diminution of pressure, and it the gases remained in suspension in the water the volume of the water would be increased.

Calculation shows that with a diminution of atmospheric pressure from 30 in. to 28 in. of mercury—that is, from 15 lb. to 14 lb. per square inch — the bubbles of gas liberated in this well 340 ft. deep would raise its level by 1.8 in. By observation, the rise of the well under such a barometric fall amounts to 8 in., and therefore the liberation of gases

was constantly higher on Sundays than on Saturdays and Mondays, and that even a public holiday was accompanied by a decisive rise in the well under observation. Mr. Dobson, Christchurch City Engineer, has informed me that the installation of a city water-supply has been followed by the breaking-out of springs in numerous places about the city, and he explains this as follows: In the early days of the city's life wells sunk on some of the higher ground had a static level of 1 ft or 2 ft. above the ground. As more and more wells were sunk to the same stratum, the static level was lowered; those on ground a foot or two lower continued to flow, but those on higher ground had their static level reduced to below that of the ground, ceased to flow, were abandoned and forgotten, and their mouths covered, up. On the installation of the city supply many users of artesian water stopped their flowing wells, the static level recovered itself, and the old abandoned wells recommenced their flow, sometimes in such inconvenient places as cellars, public parks, and important streets. The explanation seems very probable, and emphasizes the interdependence of wells sunk to the same stratum. Mr. Dobson further informs me that he on one occasion fitted a pump to a particular flowing well, and started to work the pump with a steam-engine, with the result that as long as the pump was at work all the wells in the neighbourhood ceased to flow. It was primarily to escape this interference of one well with others in its neighbourhood that I commenced observations on the comparatively isolated well at Lincoln, and it was the evening rise that was the original object of the inquiry. As stated before, there are only four other wells of the same depth as the College well within a radius of two miles; the nearest of these is three-quarters of a mile away, and I felt that I could secure from the owners of all these wells any co-operation necessary for my observations.

The object for which the investigation was undertaken has, however, not been accomplished, since no light has been thrown on the evening rise, except that it does exist, and that it is not caused by the shutting-off of neighbouring wells. Out of the fifty-one weeks during which the observations have been made, the weekly averages of the evening readings have been higher than those of the morning readings on thirty-six weeks, equal on four weeks, and lower on eleven weeks. The following table shows the averages of all the readings of each month, with the evening rise:—

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

The evening rise is thus fairly well marked. During the months of October and November I personally secured that all the wells in the neighbourhood were running continuously, with the exception of one (three-quarters of a mile away) which its owner was good enough to shut off from 7 to 9 a.m. and 4 to 6 p.m. every day. The readings during these two months were taken exactly at 8.30 a.m. and 5.30 p.m., so that the well had an hour and a half to recover any disturbance that might have been set up by the well whose flow was intermittent. That this intermittently flowing well could have any effect on the College well, so far away, is questionable, and, in any case, it was not (even during the months I did not keep special control of it) usually running in the morning or usually shut off in the afternoon. The interference of neighbouring wells may therefore be rejected as a cause of the evening rise.

Any constant variations in temperature are similarly to be rejected. I kept a record of the temperature of the flowing water just as it emerged from the ground from the 10th to the 30th October. The temperature varied from 12.81° C. to 12.90° C., and this variation was more probably due to the effect of the air on the stem of the thermometer than that of the water on its bulb. In any case, the temperature never showed any disposition to be regularly higher in the evening than in the morning, and, if it had, a much greater rise of temperature would have been needed to cause sufficient expansion of the water (inside an iron pipe, on which the scale was carried) to account for the observed rise in the static level. The water in the gauge-glass is, however, practically the same water all the time, and therefore takes on to a considerable degree the temperature of the atmosphere. It varied from 10.0° C. to 23.9° C. during the month of October. The higher readings were, however, on all but three occasions obtained in the morning, owing to the sun shining on the gauge-glass and above-ground portion of the well-tube in the morning and not in the evening; indeed, the highest reading (23.9° C.) was obtained in the morning, and on the same evening the temperature was 12.0° C. In any case, an average evening rise of temperature of about 25° C. would be needed to cause a 4 ft. column of water (in a glass tube with an independent scale) to expand sufficiently to account for the observed rise in the static level. A shrinkage of the wooden scale in the evening would also explain the rise; but means to detect and measure this, if it occurred, were not at hand, and the line of investigation held little promise. During the months, of October and November, also, records were kept of the barometric pressure in the mornings and evenings, and it was found that the readings were, on the average, lower in the evenings than the mornings. The amount of the decrease in the barometric height in the evenings was 0.07 in., sufficient to account for a rise in the well of 0.56 cm., or more than the actually observed rise. The barometric observations were, however, taken on an aneroid barograph, the mercurial barometer unfortunately being out of repair. There is, I suppose, no question that the temperature of a living-room is higher in the evenings than the mornings, and I have rather good proof that increase of temperature depresses the record made by an aneroid barograph. The apparent fall of the barometer each evening is, therefore, only a temperature effect, and

manner and to the same degree as in the Lincoln well, but there is a much less decline in its static level during a similar period of almost similar rain. (3.) There is no sign of floods in the Waimakariri influencing the well. (4.) There is no sign of agreement between the graph of the well and that of the barometer, however the latter is manipulated. (5.) There is an evening rise. Points (2) and (3) are illustrated by the following graph (fig. 8), which is comparable to fig. 2, both graphs being on the same scale.

Fig. 8.—Weekly Averages of Height of Museum Will And Weekly Totals of Rainfall at Lincoln (1894–95).

The want of agreement between the graph of the well and that of the barometer may be explained either as the result of the Waimakariri assisting the rainfall to supply the well, or as the result of the interference of neighbouring wells. That such interference does take place has been shown in a previous section.

The want of pronounced decline during a comparatively rainless period, and the smallness of the annual variation (10 ½ in.), opens up seriously the question as to whether the Waimakariri does assist the supply of the flowing wells in Christchurch. In favour of the rainfall being the sole source of supply are the following facts: (1) The rise of the well after rain; (2) the absence of effect of even the greatest floods on the river (see fig. 8, 1st December); (3) the diminution of the static level of the wells as each additional well is put down. The Museum well has fallen 4 ½ ft. in fifteen years, and there is a generally expressed opinion that all the wells in the town are similarly affected. This would be the natural effect if there were a restricted supply of water, such as a rainfall of 25 in. affords. If the lowering of the static level of the wells is an indication of the lowering of “the water in the water-table at the outcrop (and it is difficult to suppose otherwise), then the wells in the town are robbing the crops in the country of the supply of water that they should receive by capillary rise, a matter of some importance on light shingly ground.