10
240 ELLIS-~~EAN TEHPERATURE OF AIR AT GREENWICH, 1849-88. ad four times in October, the greatest deviation on any day in these months being respectively 1O.8, 1O.4, 1O.4, and 2O.6. The average deviation taken without regard to‘ sign is respectively OO.4, OO.5, OO.6, and 00.6. The differences are evidently of accidental character, as is shown by the agree- ment of the monthly means. Thus the monthly means of the daily values in the Registrar General’s report for January, April, July, and October 1888, are respectively 37’-8, 43O.4, 57O.9, and 46O.0, the monthly means of the values afterwards given in the Greenwich volume being respectively 97O.9, 43O.5, 58O.0, and 46O-0, which are practically the same. And in the whole eleven years, 1878 to 1888, there is no instance of a deviation greater than 0e04 between the corresponding monthly means. I think it will be seen from this comparison, that the values, given in advance in the Registrar General’s report, are fairly representative and if required may be taken as representative of those, afterwards found from the photographs, which appear in the annual Greenwich volume, I do not pretend to say that the corrections for Greenwich would give equally satisfactory results for other places, but it is probable that they could be usefully employed for the-com- parison of results for different places at which the hours of observation were not the brame. On the Comparison of Thermometrioal Observations made in a Stevenson Screen with CorrespondingObservations made on the RevolvingStand at the Royal Observatory, Greenwich. BY WILLUM ELLIS, F.R.A.S. [Received May 11th. Read Jnne 17th, 1891.1 TECE use of the Stevenson screen for exposure of thermometers for determina- tion of air temperature having become so general in England during late years, the Royal Meteorological Society, in the gear 1883, appointed a Com- mittee, of which I was myself a member, to consider the best pattern of screen for use by observers in connection with the Society. Their report, in which the adopted form of screen is described, will be found in the Qunrtevly Jouinnl, Vol. X., page 92. Afterwards, in the year 1886, on my sugges- tion, a Stevenson screen of the pattern recommended was set up at the Royal Observatory, Greenwich, adjacent to the open revolving stand, commonly known as the ‘( Glaisher Stand,” for the purpose of comparing the indica- tions of thermometers placed in the screen with those of the corresponding thermometers of the revolving stnnd. A description of the revolving stand will be found in the Introduction to the Greenwich annual volume. The screen

On the comparison of thermometrioal observations made in a Stevenson screen with corresponding observations made on the revolving stand at the royal observatory, Greenwich

Embed Size (px)

Citation preview

Page 1: On the comparison of thermometrioal observations made in a Stevenson screen with corresponding observations made on the revolving stand at the royal observatory, Greenwich

240 ELLIS-~~EAN TEHPERATURE OF AIR AT GREENWICH, 1849-88.

a d four times in October, the greatest deviation on any day in these months being respectively 1 O . 8 , 1O.4, 1O.4, and 2O.6. The average deviation taken without regard to‘ sign is respectively OO.4 , O O . 5 , O O . 6 , and 00.6. The differences are evidently of accidental character, as is shown by the agree- ment of the monthly means. Thus the monthly means of the daily values in the Registrar General’s report for January, April, July, and October 1888, are respectively 37’-8, 43O.4, 5 7 O . 9 , and 46O.0, the monthly means of the values afterwards given in the Greenwich volume being respectively 97O.9, 43O.5, 58O.0, and 46O-0, which are practically the same. And in the whole eleven years, 1878 to 1888, there is no instance of a deviation greater than 0e04 between the corresponding monthly means. I think it will be seen from this comparison, that the values, given in advance in the Registrar General’s report, are fairly representative and if required may be taken as representative of those, afterwards found from the photographs, which appear in the annual Greenwich volume, I do not pretend to say that the corrections for Greenwich would give equally satisfactory results for other places, but it is probable that they could be usefully employed for the-com- parison of results for different places at which the hours of observation were not the brame.

On the Comparison o f Thermometrioal Observations made in a Stevenson Screen with Corresponding Observations made on the Revolving Stand at the Royal Observatory, Greenwich.

BY WILLUM ELLIS, F.R.A.S.

[Received May 11th. Read Jnne 17th, 1891.1

TECE use of the Stevenson screen for exposure of thermometers for determina- tion of air temperature having become so general in England during late years, the Royal Meteorological Society, in the gear 1883, appointed a Com- mittee, of which I was myself a member, to consider the best pattern of screen for use by observers in connection with the Society. Their report, in which the adopted form of screen is described, will be found in the Qunrtevly Jouinnl, Vol. X., page 92. Afterwards, in the year 1886, on my sugges- tion, a Stevenson screen of the pattern recommended was set up at the Royal Observatory, Greenwich, adjacent to the open revolving stand, commonly known as the ‘( Glaisher Stand,” for the purpose of comparing the indica- tions of thermometers placed in the screen with those of the corresponding thermometers of the revolving stnnd. A description of the revolving stand will be found in the Introduction to the Greenwich annual volume. The screen

Page 2: On the comparison of thermometrioal observations made in a Stevenson screen with corresponding observations made on the revolving stand at the royal observatory, Greenwich

ELLIS-COMPARISON OP THERYONETER OBSERVATIONS AT GREENWICH. 241

contains a maximum thermometer, a minimum thermometer, a dry bulb thermometer, and& wet bulb thermometer, and since the beginning of the year 1887 observations of these thermometers have been regularly made at the times at which the revolving stand thermometers were read, that is to say, the daily maximum and minimum have been observed, and readings of the dry bulb and wet bulb thermometers taken at 9 a.m., noon, 3 p.m., and 9 p.m., excepting that readings of the screen thermometers have not been taken on Sundays, but since the question is one of differences, this does not signify. The results at present available are those for the years 1887, 1888, and 1889 ; they are given in complete detail in the Greenwich annual volumes, but it seems desirable now to collect them in order to show what general conclusions may be drawn therefrom. The mean differences for the same month in different years are in such close agreement that it seems quite unnecessary to give in this paper the separate monthly differences, but only the mean values for each month as found from the observations made during the three years mentioned. It should be explained, as respects the observations made at stated hours, that the order of observing the screen and revolving stand thermometers was reversed on alternate days, in order to avoid the introduction of any constant difference due to diurnal change.

Considering the numbers of Table I. it will be seen that the maximum readings in the Stevenson screen are lower than those of the revolving stand in all months of the year, not much lower in winter, but considerably lower during the summer months, also that the minimum readings in the screen are distinctly higher than those of the revolving stand throughout the year, the difference being somewhat greater during the summer and autumn than at other times. The mean of the maximum and minimum readings in the screen differs little in the winter months from that of the revolving stand mean, but is lower in other months of the year, the difference being greatest in the summer months, depending on the greater difference of the maximum read- ings at that period of the year.

The readings of the dry bulb thermometers of the Stevenson screen and revolving stand, as taken at definite hours, are in much closer agreement than are those of the self-registering thermometers ; the screen readings being at noon and at 3 p.m., even in summer, only a few tenths of a degree below those of the revolving stand. This is in striking contrast with the larger dif- ferences existing in summer, between the readings of the maximum thermo- meters of the screen and revolving stand. In the latter part of the year the screen readings at 3 p.m. are higher than those of the revolving stand, the values for the separate months in each of the three years being in close agree- ment in this respect. At 9 a.m. the differences are in most months small ; at 9 p.m. the screen readings are higher throughout the year.

The differences between the readings of the Stevenson screen and revolving stand wet bulb thermometers, at the stated hours of observation, are in close accord with the differences between the corresponding dry bulb readings, being usually of the same sign and somewhat less in amounh ; indicating that,

Theae are contained in Table I.

Page 3: On the comparison of thermometrioal observations made in a Stevenson screen with corresponding observations made on the revolving stand at the royal observatory, Greenwich

242 ELLIS-COMPARISON OF TJ~ERMOMITER OBSERVATIONS AT GREENWICH.

TABLE I.

g a.m.

-- G o 3

woo -017 -010 -013 -0.13 -ozo -013 -007 0'00 woo

-003

-013

--

-0.04

-- -0.08

~~

Excess of Stevenson Screen Readings over Readings on the Revolving Stand, from observations made during the years 1887, 1888 and 1889. I

Noon. 3 p.m. g p.m

- -- -- 0 0 0

-0.03 +003 $0.1 -0'07 +0*03 +O*I - 0 2 0 -O'IO +O.O -013 -0.07 $0.2 --0'27 -013 + 0 2 -030 -017 $0.0 -017 -0'13 +O.I - 0 2 0 -013 +O'I +0*03 $0.03 +WI 0.00 +0.07 +O'Z

4 0 7 $0.10 +O*I - 0 - q $0'13 +O*I __-- -- --

-0.17 -0.10 $0.1

-- -- --

-0.07 3-0.04 +O*I

---- -- -0'12 -0.03 $0.1

1 Sd-Registering Thermometers.

February .. Karch . . . . npril . . . . . . Kay , . . . . . rune . . . . . . rniy .. .. .. September.. lctober . . . . lJovember . . December , ,

Summer

h@St ....

-0.43 +030-00f -087 $037 - 0 2 5 -1'63 $0'37 -062 -1.77 $0'37 -07c -z03 +0'431--0.8~ -2.23 4-047 -0%

-1.17 $0'57 - 0 . 3 ~ -87 $0'57,-0'1f -0.47 +0'37,-o'o: -0.13 $027i+00;

-1'97 +0'53--0'72

------ I / 3eptember

-0.03 -0'10 -023 -0.20

Dry Bulb Thermometer.

-017 -0.07 $0.1~ -0.23 -0'17 +WIG -0 .30 -020 $ o m -0.47, -037 +O'2:

-_j_/__/__ -0.09 -0.19 -0'11 + 0 1

on the whole, the revolving stand dry and wet bulb readings and the screen dry and wet bulb readings would independently give much the same values of dew point and of humidity.

If for'the dry bulb and wet bulb thermometers we take the mean of the differences of the screen and revolving stand temperatures at 9 a.m., noon, and 3 p.m. to represent the day condition, and the 9 p.m. difference to indicate the night condition, we obtain the numbers given in Table II.

The maximum temperature in the screen is lower than that of the revolving stand, in summer much lower, and the minimum temperature is higher ; whilst the readings of the screen a d re- volving stand dry bulbs, and of the screen and revolving stand wet bulbs, as taken at stated hours, show differences of a very much less marked character ; SO much so, that my ordinary combination of observations of the screen dry bulb would give a mean temperature practically similar to a like combination of observations of the revolving stand dry bulb.

It has been ailhued that the readings on the revolving staad by day in gwnmer are unduly influenced by radiation from the ground and from the

Briefly, then, we have these results.

Page 4: On the comparison of thermometrioal observations made in a Stevenson screen with corresponding observations made on the revolving stand at the royal observatory, Greenwich

~LLIS-COMPARISON OF ?HERJIOMETER OBSERVATIONS AT GREENWICH. 243

-___ January . . .. February , , . . March ...... April . . , , . . . . May .. ,. .... June .... .... July .. .... . . Augnst .... ,. September. , . . October , . . , . . November . . . . December . . . .

TABLE 11.

By Day. By Night.

-:*03 +:'I3 -0'09 +O.IO -0.17 s0.10 -0.24 $0.20

-0'35 $0'27 -0.37 +0'17 -0.36 +0.17 -015 +0.23 +0.04 fo.20 +0.06 $023 +004 fo.13 +0.03 +0.17

-----

Escess of Stevenson Screen Readings over Readings on the Revol- ving Stand, from observations made during the years 1887, 1888 and 1889.

Month.

------ I Summer Mean, 1 -0'24

April to Sept.

Winter -~----

I Dry Bulb Thermometer.

+on21

Mean, 1 Oct. to March

Mean.

.-0'03 I +oar$

+:.05 0'00

--0'02

-0'10

-0.03

-0.04

-0.09 +0'04 +WIZ $0'14 +0*09 +OTO --_ -0'02

+0*06

+o-oz

Wet Bulb Thermometer.

By Day. --- 0

-0'01 4 ' 0 I

-0.10

-0.30

- 016 -0'18

-0'17 -0.15

WOO $0.03 +o.or +o.or

-0.13

-0'02

-0.08

3y Night.

$013

$0'14

Mean. --- +:.05 +0.06 -0.04

+0'03 4 . 0 6 -0.03 +O.OI $0.06 $0.11

$0.07 +0.07

$0-08

$0.01

white buildings i; the meteorological court, and the comparison of the revolv- ing stand maxima with the Stevenson screen maxima seems to give support to such idea. But if this be the explanation of the higher revolving stand maxima, the dry bulb readings of the revolving stand and screen at noon and at 3 p.m. should shorn similar differences, instead of differences which are in every way so very much smaller, and which, indeed, in some months are persistently reversed in direction, the screen readings being the higher. There is nothing in the positions which the various thermometers occnpy on the revolving stand which should cause anomalies. The stand is 4 ft. 6 ins. wide, the bulbs of the different thermometers are all placed towards the centre of the stand, which is kept with its inclined side always towards the sun, whether the sky be clear or cloudy, being turned through a certain angle at stated times every day.

We have now, however, some direct evidence bearing upon the question of radiation. As regards radiation from the gronnd, it may be mentioned that it was because of a suggestion that ground radiation might affect the read- ings of the thermometers on the revolving stand that in the gear 1878 a horizontal circular board, 3 ft. in diameter, was fixed on the post carrying the thermometer frame, in aposition below the frame, at a height of 2 ft. 6 ins. above the g T o L d , with the object of affording protection to the thermometers iu this respect. I n the summer of the year 1886, some experiments were made to ascertain whether the removal of bhe board produced any difference

Page 5: On the comparison of thermometrioal observations made in a Stevenson screen with corresponding observations made on the revolving stand at the royal observatory, Greenwich

244 ELLIS-COMPARISON OF THERMOMETER OBSERVATIONS AT G R E E N W I ~ .

of reading. On four days of high temperature and bright sunshine, observa- tions were made of the dry bulb and wet bulb thermometers, at short intervals, with the circular board alternately removed and attached, the details of which observations are given at the.end of the Introduction to the Greenwich Mog- netical and Meteorological Observatiolzs for the year 1887. All the observations made are published, and the following are the results :-

Mean reading of dry Excese Duration Namber bulb thermometer with

18%. Experiment. Comparisons. with ciroular board of of on revolving stand circular

board removed. removed. attached.

h. m. 0 0 0

August 21 2.40 5 71.5 71.9 -0.4 ,, 28 3.30 10 71.0 70.9 +0*1 ,, 30 5.30 16 83.5 83.9 -0.4 1, 31 3.50 11 85.7 85.6 + 0.1

0 0 0

Means 77.92 78.07 -0.15 The concluded mean reading of the dry bulb thermometer with the circular

board removed is thus slightly less than with the board attached. In the case of the wet bulb thermometer, the corresponding differences were :-0°-2, +Oo*l , O O . 0 , and +0°*1, the mean difference being Oo*OO. This indicates that the removal of the circular board produces no real difference in the readings.

As respects radiation from the surrounding white buildings, the erection of the Stevenson screen afforded an excellent opportunity of testing this particu- lar question. The screen is placed a little b the east of the revolving stand, both occupying positions distant somewhat more than 30 ft. sonth of the Magnet House, which building the Stevenson screen faces ; so that, on open- ing the front vertical door, the screen thermometers become exposed to the direct influence of any radiation from the building equally with the ther- mometers on the revolving stand ; an influence which, on closing the screen door, becomes, for the sereen thermometers, completely shut off. On four days of high temperature and bright sunshine in the summer of the year 1887 observations were made of the dry bulb and wet bulb thermometers, at short intervals, with the door of the screen alternately open and shut. All details are given in the Greenwich volume for 1887, and a11 the observations made are published.

Mean reading of the dry Excese Duration Number bulb thermometer of with

of of the Stevenson Screen door open 1557. Experiment. Comparisons. with the door of

the screen open. shut.

h. m. July 4 3.16 6 8791 8795 -$4 August 3 1.10 2 69-7 69.5 +O-2

,, 4 1.40 3 70.9 71.3 -0 4 f l 6 3.40 7 83.1 a m $0.2

0 0 0

Means 77.70 77.80 -0.10

Page 6: On the comparison of thermometrioal observations made in a Stevenson screen with corresponding observations made on the revolving stand at the royal observatory, Greenwich

ELLIS-COMPARISON OF THERIIONETER OBSERVATIONS AT GREENWICH. 245

The concluded mean is thus, in an insignificant degree, lower with the door open than with the door shut. In the case of the wet bulb thermometer the corresponding differences were -0O.2, Oo*O,Oo.O, and -0O.3, the mean cliffe- rence being - OO.12. Correspondingreadings of the dry bulb thermometer of the revolving stand were also taken, the mean being 78O*75, or 1O.05 higher than that of the screen dry bulb 77O.70, no very remarkable difference at a tem- perature so extreme.

Further, when the new thermograph was set up on the ground south of the meteorological court, known as the south ground, the thermometer bulbs were carefully protected from all possible radiation effects by two boards on the north side and two on the south side, an east-end board, a west-end board, and one horizontal board below, but with free circulation of air between all the boards. On five days of high temperature and bright sunshine during the summer of the pear 1886 the north and south boards were alternately re- moved and attached at short intervals, and the photographic record of the dry bulb thermometer independently tabulated by two different observers, whose tabulations were practically identical. All details are to be found in the Greenwich volume for 1887, and, as before, no results have been omitted.

Mean reading of the dry Duration Number bulb thermometer of Excess

of of the new thermo- dth boards 1886. Experiment. Compaiisons. graph for inter- removed.

vals during which the north and

south boards were

h.m. attached. removed.

July 5 2.20 3 7897 780.8 $61 9 , 7 2.20 3 81.9 81.7 -0.2 2 , 20 4.20 6 70.7 70.7 0.0 9 1 21 5. 0 7 80.9 81.1 $0.2

Sept. 1 5.15 10 80.5 80-6 + O - 1 0 0 0

Means 78.54 78-58 +0.04 With the boards removed, the thermometer was open to any radiation from

the surrounding wooden buildings and fences. Practically the removal of the boards procluced no effect, and, indeed, before the commencement of regular work with the new thermograph on January lst, 1887, one north board and both south boards were permanently removed.

The esperiinents made with the circular board of the revolving stand show that its removal produced no real difference in the thermometer readings. The experiinents made with the Stevenson screen show that the screen read- ings are practically similar both with the door of the screen open and: shut, ancl in the experiments with the thermograph the removal of the protecting boards produced no r ed change of reading. Further, these experiments receire corroboration from the near agreement of the revolving stand readings with the screen readings at the stated hours of observation. I n such widely

Page 7: On the comparison of thermometrioal observations made in a Stevenson screen with corresponding observations made on the revolving stand at the royal observatory, Greenwich

246 ELLIS-COMPARISON OF TE~ERMOMETER OBSERVATIONS AT OREENIVIBB.

different conditions of exposure, indeed, the differences really observed (see second and third divisions of Table I.) are such as may reasonably be sup- posed to be in part due to contraction of the range in the screen, acting to depress the screen readings by day and raise them by night, and so far the revolving stand thermometers do not appear to be influenced by radiation in the way that has been alleged. I t would seem that, for the circumstance that the maximum readings on the revolving stand are so much higher in summer than those of the screen, as compared with the lesser differences between the readings taken at stated hours, some other explanation must now be sought. It is to be remembered that on a fine day in summer the temperature is fre- quently subject to fluctuations of brief duration. The maximum is, of course, the highest point touched, no matter for how short a time such temperature is maintained, and the screen maximum thermometer may be less sensitive to such changes than is the more completely exposed revolving stand maximum. The readings taken at stated times, on the other hand, are as likely to fall at the base as at the crest of such fluctuations. But even in the absence of any sufficient explanation of this apparent discordance, the observed facts, con- sidered as a whole, do not appear to at all warrant the conclusion that, because the revolving stand maxima are higher than the screen maxima, they are on that account necessarily wrong.

It has been suggested that, after having carried on for a few years com- parisons of the revolving stand and Stevenson screen thermometers, the use of the revolving stand should be discontinued. That would rather be to re- place an imperfect stand by an imperfect screen. Each may have its faults. But it seems every way better to let stand and screen be both at present main- tained, especially mhen there is a willingness to carry on duplicate observa- tions, rather than to contemplate the immediate interruption of a long series of observations made in one definite way, and commenced long before the Stevenson screen was thought of.

The mean daily and monthly values of air temperature, as given in the annual volume have, since the year. 1877, been formedfrom hourly measures of the photographs reduced to the dry bulb of the revolving stand. Table II. of this paper shows.that values so obtained are practically such as would be found were the photographic values reduced to the screon dry bulb, instead of to the revolving stand dry bulb, that is to say, the concluded tempera- tures would differ only in an insignificant degree. Such being the case, it is a further argument for retaining at present the existing system, rather than replace the revolving stand by another form, which would introduce into mean values a small but doubtful correction, as regards absolute truth, but yet one that, in making any fundamental change, should properly be taken acconnt of, although otherwise scarcely worth consideration.

Finally, it should be mentioned that all thermometers are carefully com- pared every year with the standard thermometer No. 515, a thermometer kindly supplied in the year 1875 to the Royal Observatory by the Kew Com- mittee of the Royal Socioty, and that all corrections for index error are rigorously applied.

Page 8: On the comparison of thermometrioal observations made in a Stevenson screen with corresponding observations made on the revolving stand at the royal observatory, Greenwich

ELLIS-COMPARISON OF THERMOMETER OBSERVATIONS AT (LPREENWICH. 247

+:'3O 4 2 3 -050

-127 -1'63

-0'53 -090 -0.47

-1'25

-1'77

-010 $0'33

-1.39

-0'11

-0'75

-- ---

ADDENDUM.

+:*IS $0923 +0*10-0'07 +013 -0.18

+025 -0.51 +o*z8 -0.67

+0'28 -0'62 +0.32 -0.25

+033 -0.07

f0 '20 -0'53

+0'32 -0'73

+035 $0'12 +0'30 +0'31 ---- -- +027 -0.5t

+0'23 +~'of

S o 2 5 - 0.2:

-- --- -- -_

It hau occurred to me that it might be interesting to add to the preceding paper a comparison 0 ) the observations of the thermometers placed on the roof of the Magnet House with the corresponding observations of the thermo- meters of the revolving stand. The roof thermometers are mounted in a louvre boarded shed or screen, so constructed as to give free circulation of air, with protection from radiation, It is open towards the north. The thermometers are a maximum thermometer, a minimiim thermometer, and a dry bulb thermometer, but no wet bulb. The bulbs of the thermometers are 4 ft. above the platform and 20 ft. above the ground. AS with the Stevenson screen thermometers, readings have not been taken on Sundays. The observations were commenced at the beginning of the year 188G, and the results for four years are now available.

-0.23 -0'20

-0'35 -0'45 -0'17 4 1 5 fo.05 f0'40 +0'40

-0'26

+ ~ 2 6

-- _-

Month.

-0'52 4 3 5 $0.22 -0'70 -0'38 $0.33

-0'42 -0.15 $0.43 -0.20 +o.oz +0.45 -015 +023 $0.42 +o-rz +0*30 +0.35 +0.30 S0.45 $0'48

-0'59 -0'1-9 +ox9

+O*OI~ +w25 $0.38

-0.95 -0.50 +0'12 -0.78 -0'37 +0'20

___-_- ---

I

WOO -WZg -0'02

January ................... February ................... March ..................... April ....................... May ....................... June ....................... July ....................... August ..................... September ................. October ..................... November ................. December ...................

+0'34

I Summer Mean-April to Sept.,

I Winter Mean-October to Marc

I Yearly Mean ...............

Excess of Roof Thermometer Readings over Readings on the Revolving Stand from ob- servations made during the years 1886, 1887, 1888, and 1889.

Self -Registering Thermometers. Dry Bulb Thermometer. --I

DISCUSSION. Mr. MAWLEY said that he had made a comparison of the various Greenwich

mean temperatures as obtained by Mr. Glaisher, Mr. Eaton, end Mr. Ellis, in the case8 of years where the records employed overlapped, and could find very little qpeement between them. H e then quoted some figures showing the divergencies of the various means. H e thonght it would have been better to have kept from the fbst to the simple means as deduced from the daily mexi- mum and minimum readings. Regarding the comparison betmeen the tempera-

Page 9: On the comparison of thermometrioal observations made in a Stevenson screen with corresponding observations made on the revolving stand at the royal observatory, Greenwich

248 DISCUSSION-COMPARISON OF THERMOMETER OBSERVATIONS AT GBEENWICH.

tures registered on the revolving stand and in the Stevenson screen, Mr. Ellis’s figures showed greater differences between the two forms of exposure than he (Mr. Mawley) had obtained from six years’ observations made with similar screens at Addiscombe. His esperience in observing with different screens had taught him that a great deal depended upon whether there was a good lawn under the thermometer screen. He could not understand why during the summer months the mean differences between the temperatures on the two screens at Greenwich should come out about four times as great in the case of the maxima as in that of the 3 p.m. readings.

Mr. SYMONS agreed with Mr. Ellis that it would be a great pity to give up the observations made on the revolving stand in favour of those m&de in the Steven- son screen, and he sincerely hoped that the old form of stand would continue in use, as it had now done service at Greenwich for so many years ; and he hoped that the authorities would be able to arrange that in future the readings of the ther- mometers in the Stevenson screen should be taken on Sundays, so that there would be available (1) a. record strictly comparable with that of past gears, and (2) a record strictly comparable with those from the inspected stations of the Royal Meteorological Society. He considered that meteorolom at the Royal Observatory WES badly treated by the authorities, and he thought it would be an advantage if a large part of the spectroscopic end astronomical work were carried on at a more suitable locality in the country, so that better provision could be made for the meteorologicd department.

Mr. C. HARDING said that the means given by Mr. Ellis were exceedingly valuable, and it was a great gain that correction had been made for the gaps which horn various causes occurred in the series of observations. He had always looked upon the Greenwich volume containing the reduced temperature obser- vations for the 20 years 1849-68 as a splendid piece of work, and R work of high scientific value which supplied a great need for the ordinmy inquirer. He had shared the opinion expressed by the Astronomer Royal in the text which accom- panies the Tables “that the whole work may be considered as a model of accuracy on a very large scale.” He was greatly surprised to notice the large differences which were shown between the values in the Greenwich volume and Mr. Ellis’s figures, cwerences which from their magnitude greatly lessen the usefihess of the Greenwich reductions. He had tabulated and taken the averrtges of the daily maximum and minimum redings at Greenwich since 1840, and on comparing the meme thus obtained with Mr. Ellis’s fipres, he found a very good agreement. He quoted certain differences which occur between the vdues in the Greenwichvolume for 20 years, mdthose given now by Mr. Ellis. In the volume for 20 years the January and February means for 1849 are respectively 4 2 O . 7 and 42O.3, whereas, according to Mr. Ellis, they should be 40O.8 and 43O.1, so that instead of February being colder than January by 0 O . 4 it is reelly warmer by2O.3. According to the Greenwich volume December 1860 is 3’3 warmer than December 1859, but Mr. Ellis’s values show it to be OO.3 colder, and this sort of cwerence occurs very frequently where opportunity is afforded for compmison. It is these differences which, being known t o exist, wiU greatly lessen the value of the Greenwich volume. N,o attempt having been made to correct for the absence of observations throughout the series of the Greenwich observations also vitiates very greatly other parts of the work-for example, in May 1863 the mean air temperature and the mean temperature of evaporation at 3 a.m. are both 44O.6, but it is seen that the dry bulb is obtained from 25 days’ observations, and the wet bdb from 20 days’ only, so that practi- cally no comparison is possible. In 1862 the July & temperatme mean is ob- tained from 27 days,,and the evaporation from 17 days. He regretted this want of judgment in the discussion, as he attached the very highest possible import- ance to theseries of observations made at Greenwich, which in many ways stood unique in its position with regard to meteorology.

Mr. ROSTRON said that he agreed with Mr. Ellis that it was an utter fallacy to take averages of short series of years. Ten years was certainly much too short B period, and he did not consider 50 years was by any mems too long. I n fact, he WM of opinion that for the purpose of meteorological averages the longer the series of years the better WEMI the average obtained. He illustrated the truth of his remark concerning averages deduced from short series of years by instancing the air temperature at Greenwich in the month of October, which, during the 12

Page 10: On the comparison of thermometrioal observations made in a Stevenson screen with corresponding observations made on the revolving stand at the royal observatory, Greenwich

DISCUSSION-COMPARISON O F THERMOYETER OBSERVATIONS AT GREENWICH. 249

years 1955-1666 both inclusive, was continually 111 ercess, but had now for several years past been in defect of the mean. The month of February, during the last 30 years, exhibits similar eccentricities.”

Mr. TRIPP thought that if it were possible to discover cycles of years from such a series of observations as those placed before them by Mr. Ellis, the trne mean shouldcontain

Mr. ELLIS said that in the table of mean temperature now presented the results depend on 24 hourly values on each day, adopting, as required, values from the eye observations for days on which the photographs were imperfect or want- ing, so that the whole period is completely represented. Results depending in this way fundamentally on hourly temperatures necessarily supersede for the period in question dl other values, and he hoped to see also published, in clue course, the corresponding daily values from which the present monthly results are formed. For those who prefer the simple mean of the masimum and mini- mum temperatures, the Greenwich volumes have already continuously supplied the necessary information. He considered it to be quite possible that at different places the differences between the readings on an open stand and in a closed screen might not be similar, and yet be equally true for each particdar case. I t seemed to him that meteorologists too often expected an accuracy and an agree- ment of results not at present attainable. In regard to the 20 years’ reductions, their value consists in the fidl complete information afforded on the diurnal inequalities of meteorological elements in relation to which the omission of a few days in some months, owing to imperfect photographs, becomes ofalittleprac- tical importance. Photographic processes, a quarter of a centnry ago, had not the certainty of those of modern times. H e did not imderstand tho siwprise ex- pressed at the monthly means of air temperature therein conthled varying &om the means of the table now presented, since a little examinationof the book would show that the monthly means collectedin Table 52 are those appearing in Tables 35 to 49, against the columns containing which means is to be found another column giving the number of days in each month on which the several means depend. He thought that the collected monthly means of Table 52 should have been corrected for omitted days, but the work was not done xinder his direc- tion, although he had to see it through the press. I n two tables which, however, he was able to add to the 20 years’ reductions, Tables 21 and 77, giving mean daily values of atmospheric presswe and of &r temperature (a 20 years’ average), he did take account of days on which photographs were wanting, as is noted m the introduction to the work. What has been said on the non-corre- spondence of days employed in the tables of air aiicl evaporation temperatures, Tables 38 to 67, is likely to produce an erroneous impression as to the extent of such non-correspondence. In 20 years there are 240 months, and oldy in 9 months does non-correspondence esist, and in these the difference exceeds two days in a month on four occasions only, June 1858, hhy 1963, July 1861 and 1962. For the remaining 232 months the hourly values of the two elements are strictly comparative. Au this is easily seen, as the tables themselves give com- plete information on these points. H e did not say that it wodd -not have been better to have avoided more completely such non-correspondence ; he mould hase preferred to have done so ; a t the same t h e it should be understood that this affected the work only in a slight degree, and could not reasonably be said to detract from the real d u e of the resislts.

due proportion of the high and low readings in eachcycle.