]\gt ; Relation between Barometric Pressure TVater-level .
165 would allow .
The flow of air was then so large that the cylinder of gauze * produced too much choking to give a convenient of the oil-gauge , and it had to be removed , so that the choking was then supplied by the tube connecti Every small piece oEven under these conditions t this nozzle .
The first reading of was of course not very low in this case , but it is seen to lie well on the horizontal portion of the curve .
The result for the nozzle of larger diameter is seen to be practically the same as that for a smaller one of the same shape .
In conclusion I wish to thank Prof. the Hou .
R. J. Strutt , F.B.S. , for advice on many points , and for the keen interest he has taken in the work tbroughout its progress .
On the Relation between rometric Pressure the -level in a Well at , Richmond .
By E. G. BILHAII , A.R.C.Sc .
, D.I.C. , B.Sc. , F.R. Met .
Soc. Communicated Napier Received October I. Introductor'y .
The investigation which forms the subject of the present communication must be regarded as forming part of a general inquiry into the mode of action of the well at Kew Observatory , Richmond .
The well was sunk during the course of some extensive alterations to the building about four years ago , and the water-level has been continuously recorded since the end of July , 1914 .
The mean level for each day expressed in centimetres above Ordnance Datum has been published in the 'Geophysical Journal , ' and the extreme values for each month have usually been given in the Monthly Weather Report .
After being in action for two years , it was thought desirable to undertake a somewhat detailed examination of the records which had accumulated .
The most interesting feature of the inquiry proved to be the periodic variations of .
level , which were to be attributed to the tidal oscillations of the neighbouring River Thames .
The most noteworthy response of the well was found in the case of the lunar fortnightly oscillation , which could , indeed , be plainly seen by merely plotting successive daily mean levels on a fairly open scale .
The solar and lunar diurnal , semi-diurnal , ter-diurnal , .
oscillations were Mr. E. G. Bilham .
Relation between Barometric sought for by forming solar and lunar diurnal inequalities at two-hour intervals and analysing the resultant variations .
During the course of this work , it soon became evident that the extent of the solar diurnal variation was very much greater than one had any right to expect , and one inferred .
that the solar tides were not alone concerned in its production .
The effect of barometric pressure suggested itself as a likely factor , and an investigation was taken in hand .
Full details of the method of registration and of standardising the traces , as well as a description of the site of the well , will appear in the paper the results of the work on the tidal phenomena .
Some of the are quoted in this paper where necessary .
With these exceptions , the inquiry to the effect of barometric pressure forms a more or less complete study in itself , and , from this point of view , the results would seem to be of interest to justify .
being brought before the Society .
The well referred to penetrates to a depth of 11 feet below the basement of the , the bottom being 150 cm .
above Ordnance Datum .
The soil consists of sand and coarse gravel sting upon London Clay , of about 9 feet .
The well is about 300 yards from the Thames .
II .
between of The direct detection of a relation between atmospheric pressure and water-level is a matter of no little difficulty , on account of the multiplicity of factors which have been found to control the latter .
Beside the ular tidal effects , the well shows the usual type of seasonal variation , with a maximum late in winter and a minimum in the late autumn .
During the winter the changes of level may be very rapid and irregular , particularly at times of flood .
The well also appears to be highly sensitive to Any connection between the height of the barometer and the of the water in the well therefore almost certain to be masked in a short sel.ies of observations by the large , deviations due to those causes .
After consideration , ths following method adopted : The change of level between 4 .
and 7 .
G.M.T. each day measured on the trace in millimetres .
The given in the Daily Weather Report for the barometric change in three hours at 7 .
was then written down in an adjacent column .
This was done for each day of the two years , August , 1914 , to July , 1916 , with only omissions as were necessitated by loss of record .
For brevity , the of level and of pressure the three hours ending at 7 .
will be denoted by and respectively ; and will denote corresponding small changes of level and pressure in general .
In the Daily Weather eport the value of takes the form of an Pressure and the Water-lerel in vatory .
167 integer\mdash ; the unit being the half-millibal.\mdash ; to which a positive or negative sign is prefixed , to indicate a rising or falling barometer .
* The same convention adopted in the value of .
On account of the variation of pressure , the mean value of SP is about , but the of depressions and anticyclones introduces sufficient variation to produce a fairly normal distribution , in which values up to mb .
occur with sufficient frequency .
Retaining the unit of the half-millibar , we then , place a viven value of in one of 19 classes , according as the value of SP iven in the Daily Weather Report is , , , , , or A few lues of SP of magnitude occurred the riod under examinatio1 ] , but they have not been employed in the subsequent wolk .
The mean value of for each .
with the fi.equency , for whole period , is given in Table I. Table I. unit 0vati .
There are some ularities , but the tendency tive values of to be ssociated with positive departures from the mean value of SU is manifest .
The values of and in the Table a correlation coefficient of , which is high enough to place the reality of the connection beyond casonable question .
The ression equation is converting the units to millibars , whence it that on the an ease of of 1 ml a eliod of three hours , is } anied 1 ) a depression of the -levcl amountin to rather more than a millimetre .
The preliminary work having been successful , in so far as it indicated definite connection between changes of essure and , it was thought desirable to extend the inquiry with the object of determining : More accurately , a negative sign indicates thnt the barometer higher at 4 .
than at 7 , and verse Mr. E. G. Bilham .
between Barometric ( 1 ) Whether the value of ratio depended upon the qign of upon the value of .
in other words , whether the relation between and were linear .
( 2 ) Whether the value of were subject to seasonal variation , ( 3 ) Whether there were any lag in the response of the water-level to changes of pressure .
III .
of bjtween and The ures given in Table I are too irregular to permit of any conclusions being drawn regarding the linearity or otherwise of the relation between and .
In order to obtain a smoother curve the fifteen classes included between and were reduced to five by combining successive groups of three .
The results obtained in this way are given in Table II .
Table II .
Mean Mean . .
( in mm. ) ainst ( fig. 1 ) it is seen that the slope of the curve decreases uniformly in from negative to positive values of .
Before accepting this conclusion as representing a genuine phenomenon there are .\mdash ; The mean values of and are shown by the large cross .
certain points .
to be considered .
In the first place , the observations are not leally homogeneous .
The more rapid changes of barometric pressure are almost entirely confined to the winter months .
Consequently , values of of and wards are very much more frequent during the winter than last entries in Table II as fairly ) eneral conditions hout the year .
Again , it will be observed that the curve does not pass through the point representing the mean values of the quantities .
This means t$ that a small depression of wate .
-level occurs without c of pressure , which seems to indicate that " " accidental\ldquo ; features have not been completely eliminated .
For the purpose of investigating the existence of a seasonal variation of Su/ p the data were subdivided into three groups in a manner to be explained presently , From what has been said above , it essential to treat the data according to a seasonal classification before further ress can be made towards interpreting the results so far obtained .
I will therefore proceed at once to the consideration of our second object of inquiry , the results of which may be expected to throw some light on the question as to whether the relation between is really linear or not .
IV .
Seasonal Variation .
It has been previously remarked that the water-level at Kew shows a pronounced seasonal variation , the highest levels normally in winter , followed by a steady drop to a minimum late in the autumn .
In the work on the tidal phenomena it was found that the sensitiveness of the water-level to all kinds of variation was intimately related to the of the water .
For example , the ranges of the solar and lunar diurnal variations obtained from monthly mean inequalities were both found to increase with rise of the mean level , the correlation coefficients about in each case .
It was therefore thought best to take the water-level as a basis of classification rather than to employ the ordinary seasonal subdivisions of the year .
The following roups of months were formed:\mdash ; Group -Level Months .
January ) 1915 February ( 370 ) 1915 March 1915 December ( 355 ) 1916 January ( 0 ) ) 1916 February March ( 400 ) 1916 April ( 364 ) Mean level .
above M.S.L. cm .
below ground level .
Mr. E. G. Bilham .
Relation between rometric Group II.\mdash ; Intermediate Months .
1915 April ( 273 ) 1915 May ( 252 ) 1916 May ( 286 ) 1916 June ) level cm .
above M.S.L. cm .
below ground level .
Group III.\mdash ; Low-level Months .
1914 August 1914 September ( 183 ) 1914 October ( 176 ) 1914 November ( 175 ) 1914 December ( 237 ) 1015 June ( 237 ) 1915 July ( 214 ) 1915 August ( 217 ) 1915 September 1915 October ( 208 ) 1915 November ( 225 ) 1916 July ( 223 ) Mean level cm .
above M.S.L. cm .
below ground level .
The figures in parenthesis yive the mean heights of the water in centimetres above mean sea level .
The above classification , although quite arbitrary , was found to be very satisfactory in practice .
There were many and apparently no serious objections attending the use of the same oups in the present work , and they were accordingly adopted .
The method followed has been similar to that used in Table II , with the exception that only three classes have been retained , comprising values of and .
The observations in each class are usually sufficiently numerous to fairly reliable values , and the range of pressure is quite large enough , in view of the ultimate object of the inquiry , to determine what portion of bhe solar diurnal variation is to be uted to the effect of barometric pressure .
In order to some confirmation of the ures by comparing results obtained under nilar but independent conditions the low-level and highlevel months have been considcred in two groups .
The complete results are given in Table III and shown graphically in fig. 2 .
Pressure and the Wate'-level in a TVell Observatory .
171 Mr. E. G. Bilham .
Relation between rometric It will be seen that with one noteworthy exception ( high level , 1915 ) the three points lie fairly on a straight line passing through the point corresponding to the mean values .
It seems fair to conclude that the exceptional character of the -level curve must be entirely due to accidental circumstances .
It must be remembered that changes of level are much more rapid and irregular during the high-level months than during other periods , and consequently a greater number of observations are necessary to reduce errors due to " " accidental\ldquo ; movements ( including in the latter term all which are not caused by changes of pressure ) to a sufficient extent .
It appears , then , that under similar conditions the effect of an increase of pressure of about , spread over three hours , is equal and opposite to that produced by an equal decrease of pressure .
Within those limits it would seem justifiable to generalise that conclusion .
From three points only it is of course not possible to obtain the true course of the curve on which they lie , but we may fairly assume that a fourth poin is given by the mean of and .
The results indicate that the four points lie on a straight line .
We may therefore write in eneral at any so long as does not exceed .
per hour .
The factor always negative but varies with the amount of water in the well .
The values of the factor , computed where necessary by the method of least squares , are given in the Table .
The nature of the correction which has been applied in order to obtain the results in the last column will be explained in the following section .
It will be seen that the sensitiveness of the waterlevel to changes of barometric pressure is conspicuously dependent upon the amount of water in the well , high levels associated with high sensitiveness .
This relation is similar to that observed in the case of lunar tides of short period and also in the case of rainfali .
It appears , then , that the height of the waier in the well may be regarded as an index of sensitiveness in general .
The only exception which I have observed occurs in the case of the lunar fortnightly tides , the effect of which upon the well does not appear to much with the level .
Although it is convenient to state the general conclusion in the above form , there is good reason for believing that the connection is purely an indirect one , due to the fact that those conditions of the soil which are necessary for high sensitiveness occur only at times of level .
The condition of the soil , indeed , appears to be the chief if not the only factor which controls the sensitiveness .
In the absence of accurate measurement a good idea of the sensitiveness is conveyed to the eye by the amount of incidental movemsnt , Pressure and the Water-level in Well at 173 or " " embroidery\ldquo ; as it is usually termed , by the trace .
Now when the water is rising rapidly from a low level to a higher one , which is frequently the case late in autumn , the appearance of the trace in all the cases I have 9 examined is very different from that during a fall of level the same range .
In the latter instance the sensitiveness appears to be much greater .
I am unable to say whether the response to each factor is equally ' affected , but in the case of barometric pressure at any rate the sensitiyeness characteristic of " " level\ldquo ; docs not appear to manifest itself until the ' : water has ceased rising for the time bei .
It is clear , then , that the 3 is decided by the condition of the soil rather than by the absolute level .
The difference between the level results for 1915 and 1916 may , therefore , to some extent be real .
Reference may be made here to fig. 4 , which represents the relation between level and sensitiveness so far as I have been able to ascertain it .
I have made no attempt to represent the curve mathematically because the number of points is too small to define its exact shape with certainty .
It would appear to be either hyperbolic or exponelltial in form .
On the first supposition the sensitiveness is inversely proportional to the depth of the water-line below some fixed level .
The extrapolation of the curve as it stands would result in the sensitiveness infinite for a level of about 430 cm .
or about .
below the sui.face .
This is , of course , out of the question , but there can hardly be any doubt thaG the value of increases rapidly as the water-level approaches the surface .
of the For the purpose of ascertaining the main features of the phenomena under investigation , the data employed would seem to have been quite satisfactory .
It will , however , ultimately be necessary to know the value of in the different groups with some accuracy .
It is desirable therefore to form some idea of the probable errors involved .
Consider first the barometric data .
The " " barometric tendency\ldquo ; is determined at Kew by means of a Dines float barograph .
The change of pressure in three hours is estimated with the help of horizontal lines , ruled by hand , whose distance apart is intended to represent 5 .of pressule ( 10 on the scale used in the Daily Weather Report ) .
As the estimate has to be made while the chart is on the drum and under conditions of illumination which leave something to be desired , there is obviously scope for errors which may be important in the present connection .
In order to form an idea of the accuracy obtained , I have compared the mean value of the barometric change from 4 .
to 7 .
in each month obtained from the the two yeals the mean alues are .
; We have now to decide whether this difference is due to some systematic error of observation .
The root mean square value of for individual lnonths is .
If this is due to casual error the standard value of for 24 months is or , and the probable error two-thirds of this , or .
The qerved value is more than three times this amount , which , seems to gest the probability of a systematic error in If we take .
as the standard of an individual hourly reading of pressure , the standard error of a monthly mean for a given hour is or , and the standard error of in a given month or .
The stalldard error of is a good deal larger than .
The fact that the estimate is only made to the nearest halfmillibar introduces a standard error of .
in an individual reading which would give rise to a standard error of .
in .
The standard error of is therefore or mb .
We have underestimated , since we have ssulned that the estimate of the tendency is rect to the nearest half-millibar .
The agreement with the observed value .
is therefore satisfactory .
One nity for the introduction of a systematic error immediately ests itself , nanlely , the ruling of the chal'ts .
The ) magnificationwtor of the instrument is , with probable error of about 1 per cent. Tho distauce apart the 5 ) .
rulings should then be or mm. oGual s. was measured on two samples of 20 selected ( ( ndont from the for he ] 91 Both } the same value , vi probable .
lt ) ears then the } on the cent. too .
If tion on the value of ) ecomcs 0 , giving ) lnl colrection is applied ) indiof it is found the erence is in ative in 9 out of is no tion the remaining thing but casual .
A correction of per cont .
applied to the results in Table From the oiug disctlssion it Inight appear that it vould have ) desirable to use more precise barometric data\mdash ; hourly tabulations for .
The relative uncertainty in the meau value of is , however , in ryeneral t that in , so there was nothing to be gained by the latter quantity with all possible ) recision .
Ths in does not arise so much of measurement as from luctuations of water-level , dus to causes other than .
of pressure , and is not easily estimated .
The best of appraising the results is to compare values tained from of months belonging to the same category .
In the -level class there is some discordance between the values for 1915 and 1916 , although the mean levels were very nearly the same .
It would appear that the uncertainty in the final may amount to 20 per cent. or more .
* The difference in the case of the two low-level groups is in the direction to ) expected from the difference of level .
The figure for the whole period rests upon about 600 obsel.vations .
appears to be uncertain to the extent of about 8 per cent. ovel the whole period .
The uncertainty in is probably of about the sams order of nitudc , so that we can hardly suppose that the probable .
is less than about mm. VI .
of to In the preceding sections we have compared the changes of water-level which occur simultaneously with given changes of pressure , and del.ived certain concIusions from the results .
It is obvious , however , that those * This is the uncertainty involved in assigning a certain value of to the levcl of 360 cm .
It necessarily includes the effect upon the average alue of real differences of sensitiveness at the same level .
VOL. XCIV.\mdash ; A. AIr .
L. con clusions 1nay ) largely in crrol .
unless il can } ) shown that the response of the well is not associated with During -leyol is difficulty in iden lnovements ol water-levcl which have taken place as tho esult of well-marked of barometric passage of a depression , for examplc , is shown by a peak on -level trac Two examples of such movements shown in fig. 3 .
In diagram time scales of the raph and water- FIG. 3.\mdash ; Broken line\mdash ; water-level trace ; full line\mdash ; barograph trace inverted .
Vertical divisions each represent 20 mm. in water-level and .
of pressure .
level recol.der have been brought into equality , and the barograph trace has been invcrted .
The former instrument records on a daily chart , and the latter a weekly , one hour .
represented by inch and inch respectively .
Time malks are made twice daily on the water-level , so that it is generally possible to obtain the time at any point on the trace to within about ten minutes .
Judging from the two examples figured , it would appear that the pressure changes are very faithfully reproduced by the water-level recorder simultane usly with their occurrence .
If there is any lag it must be a matter of minutes only .
In order to verify this conclusion I have measured the of occurrence of corresponding movements shown on the two traces in all cases that were suitable for the purpose during year 1915 .
The chief Pressure the -level in ) $i qualifications for suitability were that the movement should be sufficiently outstanding to prevent mistake in identifying it on the water-level trace , and that the feature to be measured\mdash ; the barometric minimum during a depression for be sufficiently abrupt to permit of accurate measurement of the time of its occurrence .
The results triven in Table , and it will be seen that they show no systematic indication of in the water-level .
Where times differ appreciably the is as often in one direction as in the other , undoubtedly due to some canse for uncertainty in the measurement .
In most , the times wit.bin the limits of the errors of measurement .
able \mdash ; Tilne of Occnrrence , etc. , and of onding Water-Level ents d 1915 .
Rem \ldquo ; , 1 inimtLll :$ ' 2 2 Iimes 1tgree if steady fall of -level is aliowed Sudden ) 8 10 10 9 for .
1 1 Presbltre nearly constant to 14 hours .
lrace- .
It mtty noted in } ) that the from which fig. 3 was drawn may be used to tain of .
Thus:\mdash ; \mdash ; Mean level cm .
; pressure range .
; water-level mm. ; whence Mr. G. Bilhalll .
\mdash ; Mean level cm .
; pressure range ; water-level ) mm. ; whence rrlnl .
Thess would to be quite reliable and I have plotted them eGher with the data of able III order to obtain a curve , giving the variation of sensitiveness with water-lcvel .
The rapid increase of with of level is striking .
WATER -LEVEL IN CMS .
ABOVE .
L. .
to the question of la , a piece of ridence may be quoted in suppor of the oonclusion arrived at .
times of wind a of the raph trace occurs as a result of the phenomenon known as " " ) nmping.\ldquo ; The water-level traces are affected in a very similar manner .
Now , if there were any lag in the response of the well , it is difficult to see how the rapid of pressure to which the movements are due could be regi.stered .
We may surlnise that any ishness in the response to ordinary ) of ) ressnre would be accompanied by insensitiveness to and rapid movements .
The records show , , that the latter are recorded in a silnilar nner apparently on the same scale as the former .
Pressure and the -level in at Kew Observatory .
179 , of Results and Conclusion .
We may summarise the results of the inquiry as follows:\mdash ; ( 1 ) At all seasons of the year the -level is sensitive to changes of .
pressure , a rise of the barometer associated with a fall of water-level and rice ( 2 ) Within certain limits the change of level is propoltional to the change of pressure it .
( 3 ) The magnitude of the of level produced by a yiven change of pressure increases rapidly as the subsoil water-level rises .
( 4 ) There is no appreciable in the response of the water-level changes of pressure .
These conclusions must be understood to refer to the level of the water in the well , because we have no justification for assuming that the effect of a of pressure upon the free surface of water is the same as its effect upon the water held in the interstices of the soil .
In fact , it is possible that the phenomena may depend upon a differential action of this nature .
The fact that barometric pressure is a factor of importance in relation to the movements of subsoil water appears to have been first recognised by .
Baldwin Latham .
* His observations related chiefly to the Bourne flow in the Croydon district , and included of deep wells .
So far as can be ascertained from the brief notices of his two papers read before the British Association , his ation of the response of the wells to pressure changes led to conclusions very similar to those enunciated above .
We may suppose , then , that the phenomenon is a general with limitations imposed by the formation\mdash ; exhibited by both deep and shallow wells .
A further result obtained by Latham was that the rate at which percolation occurred increased with diminution of pressure .
This conclusion be confirmed by direct observation with a , and attributed the phenomenon to the action of dissolved air and other gases in the water .
He supposed that a diminution of pressure would result in the liberation of a certain amount of the gases , and that the consequent increased aeration of the soil would facilitate percolation .
An increase of pressure would reduce percolation by causing an absorption of gases in the soil by the water .
The chief difficulty in applying Latham 's theory to the Kew well appears to me to lie in the fact that there is no lag in the response of the well to changes of pressure .
All the evidence which has accrued from the investigation of the movements of subsoil water at Kew indicates that percolation can On the Influence of Barometric Pressure on the arge of Water ings , \ldquo ; ' B.A. Report , ' 1881 , p. 614 ; 1883 , p. ISlr .
E. .
Bilham .
Relation between netrir , only operate slowly in bringing about changes of level .
Even during times of flood , when conditions are most favonrable for rapid penetration , .
the water-level in the well continues to rise for hours after the surface water has begun to Qide .
It is therefore difficult to believe that any effect upon some change in the rate of percolation could follow so promptly npon the initiation of that change as we have found to be the case in the ] ) henomenon mder notice .
There appears , therefore , to be a strong probability that the effects of pressure upon well-level are not secondary effects due to changes in the rate of percolation .
I have alread indicated an alternative direction in which to look for an explanation of the observed results , namely , a differential action of pressure upon the free surface of water in the well and upon the subsoil water .
The well behayes very much like the open arm of a manometer connected to a reservoir of air .
If we suppose that the air contained in the interstices of the soil immediately above tho water-line , or in cavities , is cut off from external pressure changes , the general character of the movements would seem to bc sufficiently well accounted for .
Any change of barometric pressure would result in a readjustment of the water-level in the soil and in the well , until uniformity of pressure at any given level below the water line was re .
ished .
If view is correct , it follows that the water-level in the well does not in general represent accurately the true subsoil water-level .
The two levels would only cid at some particular height of the barometer .
It must also be supposed any volumes of air concerned in the action are very pel.fectly cut off from external pressure changes , otherwise the well would behave like a micro-barograph , responding only to sudden changes of pressure .
The results , however , point to the conclusion that the rate at which the pressure is ' is of no consequence .
old appear at first sight that this postulate is in opposition to the well-known phenomenon of the emission and absorption of gases by the soil in response to barometric changes .
In deep mines this effect is of so great importance that special precautions may be necessary to preyent explosions mder certain barometric conditions .
In the case we are with , however , the local circumstances are rather peculiar .
The well is sunk through a concrete floor , is lined with glazed pipes , and is surrounded by a system of brickwork vaults .
Under these circumstances it seems quite likely that cavities may occur containing imprisoned air in sufficient quantity to produce the observed results .
effects observed by Latham may also have been produced in some such way .
A third possibility is that the phenomenon may be the result of the dist , option of the earth 's crust and the water-plane under variations of pressure .
Pressure the -level in Well at Obserratory .
Sir G. H. Darwin*in 1882 showed that between two places on au otherwise flat earth of the rigidity of glass a difference of level of about 9 cm .
would exist as the result of a difference of pressure equivalent to 1- ) cm .
of Over the free ocean the water would be depressed cm .
for each centimetre of mercury increase of pressure .
If the subterranean water plane were distorted to the same extent the depression per centimetre of mel.cnry measured by an instrument on the surface would be 13 cm .
or which corresponds to a value of .
The actual chances of level measured at Kew were considerably smaller than this , which implies either that the rigidity of the earth is much less than that assumed by Darwin or that the distortion Of the subsoil water surface is a good deal less than over the free ocean .
Darwin concluded that the elastic compression of the ground must take place without sensible .
delay , but it is at least doubtful whether the same statement is applicable to the subsoil vater .
Again , it is not easy to see why the observed etfect should depend to so marked an extent upon the condition of the soil .
The above calculation may perhaps serve to some rough of the order of magnitude of the effects which might be observed under certain circumstances , but the actual litions are radically diffel.ent from postulated in Darwin 's problem .
The effects of various distributions of upon the earth 's surface have been worked out by Dr. C. Chree and others in recent years , but so as I know the solution the case in we have porous matter permeated with liquid up to a certain level is not yet ayailable .
It seems quite likely that the elfects of elastic compression ) be at least partially concerned in producing the results which I have detailed , and a mathematical treatment of the subject would be Yreat interest . .
H. Darwin , " " On the Mechanical Effects of sure on the Earth 's Surface ' B.A. Report , ' 1882 , p. 106 . .
Chree , " " Applications of Physics and Mathematics to Seismology ' Phil. Mag. , 1897 , p. 173 , et seq. VOL. XCIV .

