QUARTERLY JOURNAL

OF THE

ROYAL METEOROLOGICAL SOCIETY

VOL.

J U L Y 1917

XLIII.]

[No. 183

T H E FORMATION OF FOG AND MIST.

BY MAJOR a. I.

TAYLOR, R.F.C., F.R.Met.Soc.

[A Lecture delivered before the Royal Mcteorological Society on March 21, 1917.1

FOGmay be defined as any condition of the atmosphere near the ground

where objects a t a distance are hiaden by small particles in the air.
The air may be filled with small particles in two ways : either they may
be so small that they would take days to fall to the ground in still air, so
that they remain close to the spot where they were originally formed ; or
they may be large enough to fall at such a rate that they would clear
the air near the ground in a few minutes, but the supply of particles
from above or bcloiv continually replaces those that have fallen t o the
ground. Fogs of this second kind include fine rain, in which the
particles are supplied from above, and sandstorms, in which the supply of
particles comes from the gronnd. I n this paper, however, I propose to
discuss only the kinds of fog in which the particles are 80 small that
they practically stay where put in the air, and move only with the
air in which they are suspended. The small particles of which these
fogs are made may be drops of water or the small solid particles of
which smoke consists. In some ways fogs consisting of smoke particles
are the simpler, because when considering the formation of such fogs we
are not so much concerned with the actual formation of the fog particles,
but rather with the atmospheric conditions which prevent them from
dispersing.

SMOKE
FOGS.
There are two ways in which the particles can get away from the
immediate neighbourhood of the chimneys from which they are discharged. They may bo carried away by a wind or they may be carried
upwards by vertical convection currents. Two conditions are necessary
for tho formation of a smoke fog: firstly, the wind velocity near the
ground must be so small that the air may collect enough smoke while
passing over the town to form a fog; and secondly, the air near the
R

241

TAYLOR-THE

FOElMATION OF FOG AND UIST

ground must be relatively cold compared with the air higher up for a
period sufficiently long to collect enough smoke to form a fog.
The amount of wind which will prevent the formation of a smoke fog
depends on the size of the town, but it is evidently very small. Suppose,
for instance, that it would take 2 4 hours for tlie air to collect enough
smoke to forin a fog if there were no wind at all and the smoke stayed
in the immediate neighbourhood of the chinineys from which it had been
discharged, then, if London is 24 miles across, a wind of only one mile
per hour could keep London free from smoke fogs, because the air would
not be over the town long enough to collect enough smoke to form the fog.
The relative coldness of the air near the ground has the effect of
stopping the formation of vertical currents which would tend to
disperse the fog upwards. A moment's consideration will show the
reason for this. Let us fix our attention on a certain mass of air which
is carried upwards ' by a vertical current from one layer, A, to a higher
layer, B. Suppose that the temperature of our mass of air was originally
the same as that of the layer A from which it originated. As it rises
the pressure decreases and the mass of air expands. The expansion
gives rise to a fall in temperature, and, as is well known, the expansion
due to the decrease in pressure with a change in altitude of 100 metres
causes a fall in temperature of 1" C. If, therefore, the temperature
gradient of the atmosphere in which our vertical current is established is
less than 1" C. per 100 metres the mass of air we are considering will,
when it gets to the upper layer, find itself in air which is warmer than
itself. I t will therefore be heavier than the surrounding air and will
consequently tend to fall back to its original level. That is to say, an
atmosphere in which the temperature gradient is less than 1" C. per 100
metres has the property of tcnding to stop the formation of vertical
currents. The smaller the temperature gradient, the greater is this
tendency. When the temperature gradient is reversed the tendency is
still further increased and the effect becomes very noticeable.
The commonest example of an inversion of temperature occurs near
the ground a t night. On clear nights the ground cools by radiation to the
sky, and the ground cools the air close above it. The air at a height of
100 or 200 feet does not cool nearly so much, and there is consequently
an inversion of temperature in the first 100 or 200 feet. This is
frequently great enough to suppress practically all the turbulence a t
a height of 100 feet above the ground, although the wind a t that height
may be as great, or greater than it was during the day.
The way in which the turbulence of the atmosphere at a height of
140 feet above the ground dies away at night is shown particularly well
in the records of the pressure tube anemometer installed on the top of
Pyestock chimney, near Farnborough, Hants. In the lower half of Fig. 1
is shown the record for May 25-26, 1916. It will be seen that during
the afternoon of the 25th the wind at the top of the chimney was'
gusty and varied from about 16 to 30 m.p.h. During the night between
2 a.m. and 6 a.m. the wind waa blowing steadily at about 15 m.p.h.,
and the difference between the maximum velocity in the gusts and the
minimum in the lulls wits only about 2 m.p.h. In tlie upper half of
Fig. 1 is shown the record of the tliermograph at Farnborough. It will
be seen that the temperature dropped during the afternoon and night

TAYLOR-THE

FORMATION OF FOG AND MIST

243

from 63" F. to 40" F. It is pretty certain, therefore, that there was an

inversion of temperature near the ground.

244

TAYLOR-THE

FORMATION OF FOG A N D MIST

Another most instructive series of examples of the effect of the

cooling of the ground on the formation of eddies occurred in the week
July 3-10, 1916. The anemometer records for the four days 5th-6th,
Gth-7th, 8th-9th, 9th-10th are shown in Fig. 2 with the corresponding
thermograph trace. It will be noticed that on all these days the(wind
speed was about 10 miles per hour both during the day and during the
night. On the three days 5th-6th, 8th-9th, and 9th-10th the wind
was gusty during the day, and practically free from gusts during the

FIQ. 2.-Anemograms

showing the absence of turbulence on the nights of July 5-8, 8-9,

The turbulence persisted throughout the night of J d y 8-7.
The
thermogrnm for the week July 3-10 is shown above.

and 9-10.

night. The auemometer trace for July 6-7, however, shows gustiness throughout the whole 24 hours, though the amount of gustiness a t
night is rather less than it is during the day.
The reason for this difference may be inferred from the record of
the thermograph for the week July 3-10, which is also shown in Fig. 2.
It will be seen that, during the evenings and nights of the 5th, 8th,
and 9th the temperature fell through 26", 25", and 17" F. On the 6th,
however, the temperature stopped falling at about 8.30 p.m. (owing to a
sheet of cloud which came over the sky), after it had fallen only 9" F.

TAYLOR-THE

FORMATION OF FOG AND MIST

245

below the maximum during tlie day. It seems, therefore, that the
temperature gradient, due to the cooling of the ground on a clear night
when the temperature of the surface air falls about 20" F., is sufficient
to suppress practically all turbulent motion in a wind of 10 m.p.h. a t a
height of 140 feet (the height of Pyestock chimney), while the temperature gradient due to cooling of 9" F. on a cloudy night is not sufficient
to suppress the turbulent motion due to the friction of a wind of about
10 m.p.h. over the ground. Unfortunately I have no measurements of
the actual temperature gradients on these occasions. I have lately fixed
a thermograph at the top of Pyestock chimney, however, and it is to be
hoped that some definite. results on this point may be obtained during
the coming summer.
Another example of the way in which a cooling of the surface will
suppress the formation of eddies is to be found a t sea. Every one who
has spent much time a t sea must have noticed that on certain days,
generally in the summer, the smoke from steamers seems to hang in
horizontal streaks which frequently remain for ten minutes or more at a
height which appears to be about 50 feet above the sea. When I first
noticed these streaks, I used to think that they were formed on days
when there was no wind a t all, and that they marked the position in the
stationary atmosphere through which the steamer's funnel had passed.
Later, however, in 1913 I had an opportunity of studying these streaks
under exceptionally favourable conditions and came to a different
conclusion. At that time I was cruising on the whaling ship Scotin
over the Banks of Newfoundland and off the coast of Labrador. I n
these regions the sea consists of water which has come down out of
Baffin Bay with the Arctic current. It is therefore exceptionally cold.
The air, on the other hand, is frequently from the West and blows off
the mainland of Canada, which is very warm during the later months of
summer. Under these conditions an inversion of temperature close to
the surface of the sea is of very frequent occurrence.
The Scotia had auxiliary steam, and when under power was normally
stoked about once every quarter of an hour. Every time fresh coal was
put into the furnace a dense cloud of black smoke was omitted for about a
minute. Fig. 3 is a photograph of one of these clouds as seen from the
stern of the ship immediately after the stoking had taken place. I t will
be seen that the smoke stretches right down to the level of the water
and up to a height of about 80 feet. The little waves shown in the
photograph are sufficient evidence that there was a light breeze blowing
a t the time, and the temperature of the sea may be judged from the
floating pieces of ice, which will also be seen in the photograph.
On watching these clouds of smoke the upper part of them could be
seen to sheer off to the right relatively to the bottom, so that after a few
minutes they mould appear as long streaks leaning over to the right.
I n Fig. 4 is shown a photograph of one of them about five minutes after
it had left the ship. Fig. 5 is a photograph of two of the streaks taken
a short time after the cloud shown in Fig 3. The upper strcak is the
smoke emitted fifteen minutes before the photograph was taken, while
the lower one, half-way between the upper streak and the horizon, is the
smoke emitted half an hour, before. I believe that tlie lower streak is
actually the same smoke as that shown in Fig. 3, but I am not quite

246

TAYLOR-THE

FORMATION OF FOG AND MIST

certain. It will be noticed that thc sea surface indicates a light or

moderate breeze.
These photographs form a striking illustration of the way in which
a reversed temperature gradient near the surface will suppress the
turbulence produced by the friction of the air over the sea, for i t i s
evident that there is practically no turbulence a t the height of the
streaks, which were measured by a sextant and found to be about 30 feet,
above the sea.
We have now seen that the conditions for the formation of a smoke
fog are, firstly, that the wind shall be so light that the atmosphere has
time to become laden with enough smoke to form a fog during its passage
over a town, and secondly, that there shall be an inversion of temperature
near the ground so that the fog shall not be dispersed vertically by
turbiilence or convection currents.
This second condition practiailly limits the formation of smoke fogs to
the winter, because it is only then that an inversion of temperature near
the ground is likely to persist for a period sufficiently long to enable enough
smoke to collect to form a fog. Tho inversion of temperature which
occurs a t night in summer does not persist through the following day.
When very light winds occur in the summer, the vertical currents in
the daytime carry the smoke into the higher layers, where the wind is
usually stronger, so that the smoke gets carried away. Occasionally,
however, the wind is very light throughout the whole of the first 6000
or 7000 feet. In that case the smoke is carried by the vertical currents
during the day up into the upper layers, but it is not carried right away.
I n this way i t forms n haze which is sometimes sufficiently opaque to
obscure the sun, though it is nowhero thick enough to be called a fog.
Very light North-east winds continuing for a period of several days will
sometimes carry this haze quite a long way from London. On one
occasion, when i t spread over Farnborough, I found that i t extended to
a height of 7000 feet, and that above that height the atmosphere was
perfectly clear. On that occasion i t was possible to see a small area of
ground immediately below the aeroplane from which the observations
were made, for the first 5000 feet of the ascent. If the same general
atmospheric conditions had occurred during the winter, i t seems probable
that the coldness of the ground would have prevented the formation of
vertical currents during the day. The smoke would, therefore, have
remained close to the ground, and a fog would have resulted.
It sometimes happens, even in winter, that a comparatively high
temperature in the ground prevents the formation of fog when all other
circumstances would lead us to predict one. An occasion of this kind
occurred on February 21, 1917, when Mr. E. V. Nownham, who had
been assisting me in these investigations, was forecasting a t the Meteorological Office. On that occasion the air over London was quite stagnant,
and practically saturated with water vapour owing. to a few days of rain.
Under these circumstances i t has been customary to predict fog in
London. Mr. Newnham, however, took account of the fact that the air
had been very warm for several days previous to the 21st. He was of
opinion, therefore, that the ground must have been warmed up enough
during the warm interval to prevent the forplation of fog. Accordingly
he forecasted that fog mould not appear, and, ns i t turned out, he was

TAYLOR-THE

FORMATION OF FOG AND MIST

247

perfectly right. This example shows clearly that the question of whether
or not fog is likely to form in the course of the night does not depend
only on the atmospheric conditions at the time of the forecast or the
general conditions forecasted for the night. It depends also, to a large
extent, on the .sequence of weather for some days previous to the
time of the forecast.
%

FOGSWHICH

CONSIST OF

WATER
DROPS,

W e now turn to the most complicated, and, a t the same time, the
most frcquent source of fogs, the condensation of water vapour in the
air. The condensation is due either to cooling or to the mixture of two
masses of air a t different temperatures and containing different amounts
of water vapour.
Before discussing the various causes which may lead to the cooling
of air containing water vapour, i,t would be well to explain why cooling
and mixture may lead to the deposition of fog. For this purpose the
diagram shown in Fig. 6 is useful. In this diagram the state of the air
a t any time is represented by a point. The abscissa, that is the horizontal distance of the point in question from the left-hand side of the
diagram, represents the temperature of the air. The ordinate, that is
the distance from the bottom of the diagram, represents the vapour
pressure of water vapour contaiiietl in the air. The vapour pressrire is
proportional to tho proportion of water vapour in the air. Any change
in the state of the air from'one condition to another is represented by a
line (not necessarily a straight line) joining the two points :which
represent the initial and the final conditions of the air. All the states
of the air intermediate between the initial and the final states are represented by points on this line.
The usefulness of the diagram in considering the condensation of
vapour arises firstly from the fact that direct cooling of a mass of air is
represented by a horizontal straight line, because cooling alters neither
the vapour pressure nor the proportion of water vapour contained in the
a i r ; and secondly from the fact that if two masses of air represented by
two points, B and C say, are mixed in such a way that no heat is lost,
the point which represents the mixture lies on the straight line joining
B and C.1 If there is a larger proportion of the air represented by B'in
the mixture, then the point representing the mixture lies nearer B than
C in the straight line BC.
A t any temperature, air can 'only contain a certain proportion of
water vapour. When i t has this proportion mixed with it, i t is said t o
be saturated. The amount of water vapour which air can hold depends
on its temperature ; the higher the temperature the greater the amount
of water vapour. All the points which represent saturated air on the
diagram lie on a curve, called the saturation curve, which is shown in
Fig. 6. Any point which lies below this curve represents unsaturated
air, while any point which lies above it represents foggy air.
Strictly speaking, the points which represent the vapour pressure of
foggy air are points on the saturation line, but since vapour pressure is

* The truth of this proposition is not obvious. A proof is given in the Appendix at the
end of this paper.

TAYLOR-THE

2.18

FORMATION OF FOG AND MIST

proportional to the proportion of water present in the air till saturation

occurs, we may consider the ordinates as representing either vapour
pressure or the proportion of water present. TWOscales have been
drawn for the ordinates in Fig. 6 so that the diagram can be used either
way.l If the latter conception be adopted, then points above the snturation curve represent foggy air. If the point E, which represents the
M.M.

"

t
0

05
r

Llo-

Fro. B.-Moisture

Diagram.

proportion of water in the mixture of two unsaturated masses of air, B

and C, lies in the part of the diagram above the saturation curve (as it
does in Fig. S), then the actual vapour pressure of the mixture is represented by the point F where the ordinate from E cuts the saturation
curve. The line EF represents the portion of the mixture which is in
the form of fog particles.1
I t should be noted that the diagr3m may be used for air at any pressure if the ordinates
represent vapour pressure, but that the diagram cnn only be used at atmospheric pressure if
the ordinates represent the proportion of wnter prosent. For other praseurea the scale of
the ordinates would hnve to be altered in proportion to the alteration in pressure.

TAYLOR-THE

FORMATION OF FOG AND AIIST

249

Suppose that the state of the air a t any time is represented by the
point P on the diagram. If the air be cooled without altering its composition, the successive states could be represented by points along the
line PD. The point D on the saturation curve is called the dew-point
of the air represented by P. The result of cooling below that point
would depend on the way in which the cooling was brought about. If
the air were to cool by radiation or by any process which abstracted
heat from the whole volume of air a t once, the state of the air would be
represented by points on the prolongttion of the line PD inwards, and
fog would be produced ; but if the cooling of the air were due to cooling
of the ground, dew might be deposited. The proportion of water vapour
in the air would then decrease, the points representing the state of the
air might lie along the saturation curve and no fog would necessarily
result.
It is important therefore to consider the process by which the cooling
may take place. I n the first place there is the possibility that the air
may radiate its heat to the sky. As a matter of fact, however, air is
such a bad radiator that the radiation has apparently very little to do
with the production of fog ; though i t is possible that the cooling due to
radiation from the fog particles after a fog has started may have the
effect of making it thicker.
A much more frequent source of condensation is the cooling which
takes place in a rising current of air owiiig to expansion. This is the
most frequent cause of clouds.
A fog on high land may be due to the forcing up of air from lowerlying land by the contours of the ground, but a fog of this nature is
practically the same thing as the lower surface of a cloud into which the
high land is projecting, and it is not proposed to discuss this type of fog
in the present paper.
The chief factor which deterrnincs the temperature of the air near
the earths surface is the temperature of the surface; and nearly all
fogs, both on land and a t sea, arc due to changes in the temperature of the
earths surface undcr the surface air ; but there is a great difference between the land and the sea in regard to the way in which this change
may take place. Out a t sea the temperature of the air seldom differs
by more than one or two degrees from that of the sea. The temperature
of the sea is only altered very slowly by heat taken from, or given to,
the atmosphere; and there is practically no daily variation in the
temperature of the sea surface. The only way in which the temperature
of the sea under a given mass of air may alter is by the motion of the air
from a place where the sm is hot to a place where i t is cold, or vice versa.
I n the case of the land, the temperature of the surface of the ground
may vary rapidly through a large range. Changes in the temperature
of a given mass of surface air are due much more to changes in the
temperature of the ground a t one place than t o movement of the air
over ground, the temperature of which varies from place to place. The
temperature of the ground depends on a great variety of different causes,
such as the amount of cloud (for clouds prevent the ground from cooling
a t night or heating in day time), the velocity of the wind and the
temperature of the ground for the last day or two.
It will be seen that the conditions which determine the formation of

250

TAYLOR-THE

FORMATION OF FOG AND MIST

fog a t sea are much simpler than those on land, though they are still
rather complicated. A knowledgo of the conditions necessary for a seafog should, therefore, assist us in discussing the formation of fog on land.

FOGAT SEA.
One of the foggiest regions on earth is the part of the Atlantic which
stretches 300 miles east of Newfoundland. I n this region the temperature of the water is very lorn, owing to the Arctic current which flows
southward out of Baffin Bay along the east coast of the American
continent. During the summer i t is surrounded on three sides by warm
regions. On the west is the American continent, while to the south
and east lie the waters of the Gulf Stream. Winds, therefore, which
blow froni the East, South, and West are liable to be cooled by thc
Arctic water of tho Banks, ancl fog is of very frequent occurrence.
During the summer of 1913, I spent six months on the whaling
ship Scotin carrying out scientific work on the Banks. Of the 806
occasions on which nieteorological observations were taken during the
voyage, thero was fog 141 times. When i t is remembered that the
voyage included the journey out and the journey home, each of which
lasted nearly three weeks, i t will bc seen that the proportion of foggy
days is very high. It would have been impossible t o examine every
case in detail to see how the fog arose, but in certain cases I succeeded
in raising a kite to cxplore the upper air, and in most of those observations I traced the causo of the fog which prevailcd at the time of the
ascent. I n every one of theso cases it turned out that the fog was due
to air blowing off warm water on to the cold water of the Banks. A
few examples mill illustrate the conditions which lead to the production
of fog on the Banks.
On July 25 I raised a kite in a light Southerly wind accompanied
by thick fog. A t the tinie of the ascent the Scolia was in latitude 46"
30' N., long. 51" 41' W. The temperatiire of the sea was 10" C. (50' F.),
and the wind was blowing straight off tho warm waters of the Gulf
Stream. The position of the Scofin and the temperature of the sea are
shown in Fig. 7. On that map the distribution of temperature in the
sea is given by means of isotherms, or lines of equal temperature.
These are drawn at the Meteorological Office by collecting together all
the observations of sea-temperature taken by steamers during an interval
of about a week. It will be seen that the isotherms of 50", 60" and
70" F. are close together in the part of the sea we are considering.
I n order to find out through how large a range of temperature the
air which was pRssing the Scotin a t the time of the kite ascent had been
cooled, i t is necessary to trace on the map the path by which the air
approached the vessel, so as to find out how much warmer the sea was
in the warmest part of the path than the sea close to the Scolia. The
path was traced out by the method introduced by Shaw and Lempfert,'
the wind observations taken by steamers in the neighbourhood being
used for the purpose.

* Meteorological Office. The Life History of Surface Air Curre,ils: d Study of the
Surface TTujectorieJ of Mcwimg A ~ T . By W.N. Shaw and R. G. K. Lempfert, M.O., No.
174. London, 1906.

TAYLOR-THE

FORMATION OF FOG AND MIST

25 1

A t the time of the ascent a t 3 p.m. on July 25, the wind was
South by West, and it had been blowing steadily from that quarter for
a t least 24 hours with an average speed of about ten miles per hour.
A t 3 p.m. on July 24, therefore, the air which was passing the S'colia
a t the time of the ascent must have been a t a point 240 iniles away in
a direction South by West. This point is shown at A in the map
(Fig. 7). In the morning and afternoon of July 24 steamers in the
neighbourhood of the point A registered a wind from South-south-west
blowing with velocity about 10 m.p.h., and so the position of the air
on July 23 can be found.

FIG.7.

I n this way the progress of the air which was examined by the kite
ascent of July 25 was traced back to J u l y 16. The path is shown
in the map (Fig. 7). On examining that map, it will be seen that on July
16 the air was in B region where the sea-temperature was about 70F.
On J u l y 1 7 and 18 it was moving Southward, till on July 18 i t was
over a part of the sea where the temperature was observed by a steamer
to be 81" F. During the night of the 18th the air remained stationary,
and on July 19 i t turned Northward towards the colder water again.
During the succeeding days it continiied to flow towards the cooler water,
till on July 2 5 i t had cooled through 30" down to 50" F.
The effect of the cooling which began on July 19 would naturally
extend a t first only to the lowest layers of the atmosphere, but gradually
i t might be expected to extend upwards as the cool air was carried into

DISTRIBUTION OF TEMPERATURE AND HUMIDITY DURING KITE ASCENT ON JULY 25'-*

TAYLOR-THE

FORblATION OF FOG AND MIST

25 3

It will be iioticed that above 700 m. the temperature falls off at a rate
of about 1" C. per 100 m. This is the rate a t which temperature falls off
when the base of the atmosphere is heated. I t is the rate a t which dry
air cools owing to expansion when it rises without communicating heat
to its surroundings. It is known as the adiabatic gradient for dry air,
and it is shown as a dotted line in each of the diagrams in Figs. 8, 10
and 12. I t appears, therefore, that the air above 700 m. was left in the
same state that it was in when it was being heated from July 16 to
18; and that the cooling effect had only extended up to a height of
700 m., or 2300 ft., in the seven days during which the cooling was
taking place. This is confirmed by the fact that if the upper part of the
temperature curve be produced, as in the dotted line in Fig. 8, so as to
represent the state of the air a t the point in latitude 30" N. where the
air turned Northwards, it cuts the base of the diagram at a point
representing a temperature of 26" C. (79" F.). That is to say, the
temperature at the base of the atmosphere, when the air examined by the
kite ascent of July 25 turned to the Northward, was 79" I?. This is in
very good agreement with the temperature of the sea in those parts
which was observed to be 81" F. on J u l j 18.
We have now seen how the cooling which produced the fog of July
25 took place, and to what height it extended ; it remains to see how
high the fog extended. Besides the thermograph, I had a hygrograph
attached to the kite, and from the readings of that instrument I obtained
the relative humidity of the air a t different levels. On the left-hand
side of Fig. 8 is a curve which shows the results. I t will be seen that the
air was completely saturated, i.e. the humidity was 100 per cent for the
first 210 m. Above that height, however, the humidity fell off up to
700 m. when it mas 80 per cent. I t appears therefore that the fog in
this case extended to a height of 210 m. or, roughly, 700 feet.
700 feet is rather high for a sen-fog of this type ; they are frequently
so shallow that the top of the mast of a ship is in bright sunlight when
the hull is in such thick fog that one cannot see more than 80 or 100
feet along the deck. Occasionally, however, the fog may extend up to
3000 feet. On one occasion I sent a kite up in a thick fog, which
proved to be 2500 feet thick.
Another case of fog occurred on May 3, 1913. The path of the
air and tho rcsulting temperature-height diagram for this fog are shown
in Figs. 9 and 10. I t will be seen that in this case the kink, or bend in
the temperature-height curve, is not due to the change from heating to
cooling of the air as the wind changes from North to South, but to
a sudden change from gradual to rapid cooling as the air crosses
the boundary of the Gulf Stream. The effect is noticeable in the
upper part of the temperature-height curve of Fig. 10, where it will
be seen that the curve is by no means parallel to the dotted line which
represents the adiabatic gradient for dry air. In this case the fog was
quite thick near the surface of the sea, but did not extend to any great
height.
Perhaps the most interesting of the fog results obtained on the
voyage of the Scotia was the fog of August 4, 1913. On that occasion
the air path (Fig. 11) indicated that the air examined in the kite ascent
of Aug. 4 had been successively cooled by blowing off the warm mainland

254

TAYLOR-THE

FORMATION OF FOG AND MIST

of Canada on to the Arctic current, warmed by blowing froin there

southwards towards the Gulf Stream, and then cooled agnin by blowing
eastwards on to the Arctic water once more. The successive stages may
be followed on the map (Fig. l l ) , and also on the temperature-height
diagram (Fig. 12), where the effects of tho three stages are shown in the
three portions of the curve. It will be seen that the cooling which begsn
when the wind clinnged to the East a t 8 a.m. on Aug. 3 had extended
to a height of 370 metres by 4 p.m. on Aug. 4. The second part of
the curve from 370 to 760 metres is due to the warming which the air
ISOTHERMS OF SEA TEMPERATURE AND PATH OF THE SURFACE AIR FOR THE ASCENTOF MAY3Y

500

400

500

FIG.9.

experienced from July 30 to Aug. 3. The third part of the temperatureheight curve from 760 to 1080 metres is due to the cooling which the air
experienced on blowing off the mainland of Canada previous to July 30.
The humidity curve (Fig. 12) shows that the fog extended to a height
of about 120 metres.
From what has been said one might be tempted to the conclusion
that fogs a t sea are invariably due to the cooling of moisture-laden air
as it passes from warmer to cooler regions of the sea. When the air is
being cooled one may expect its temperature t o be higher than that of
the sea. One would expect fogs to disappear as soon as the air began
to be heated by blowing over warmer water. Under these circumstances

TAYLOR-THE

FORMATION OF FOG AND MIST

265

the temperature of the air might be expected to be lower than that of

the water. On looking through all the cases in which fog occurred
during the voyage of the Scotin in 1913, I find that there were several,
10 per cent in fact, in which the sea temperature waa higher than the
air temperature. I n Table I. is shown a comparison between the
frequencies of the occiirretice of a positive and a negative difference
between air and water temperatures during fog, and in the ahsence of
fog. The fact that air temperature is greater than sea temperature in
T E M P E R A T U R E - H E I G H T CURVE AT 1IA.M MAY 3E0

900

800-

700

u)

r~

600-

a
k

ul

Z 500-

2 400-

c3
rrl

T 300200

loo-

TEMPERATURE

O C

The position ofthe arrow in the temperature height diagram

represents the temperature o f t h e sea
FIQ. 10.

80.2 per cent of the observations taken during a fog as compared with
59.6 per cent of the observations taken in clear weather, aeems to

indicate that there is a connection between fog and excess of air temperature over sea temperature ; but there seems to be no connection between
a negative value for air temperature minus sea temperature and absence
of fog.
The fact that in clear weather there are nearly three times as many
cases in wliich the temperature of the air is equal to that of bhe sea as
there are cases in which the sea is appreciably warmer than the air, while

TAYLOR-THE

256

FORblATION OF FOG AND MIST

in foggy weather there are an equal number of cases, leads to the conclusion that possibly cold air blowing over warmer water may also give rise
ISOTHERMS OF SEA TEMPERATURE AND PATH OF THE SURFACE AIR FOR THE ASCENT OF AUG.4TH

60N

to fog. The results of the kite ascents made from the Scolia give1us no
indication of any such cause for fogs, but this is not surprising, for in
DISTRIBUTION

OF TEMPERATURE AND HUMIDITY AT 7 P.M. AUG.4T~1913

I100
1000
900
v)

800

700
5

600

g 5O0
2
W

400

300
200
I00

RELATIVE HUMIDITY (PER CENT

TEMPERATURE O C

The position o f the arrow in the temperature-height

diagram represents the temperature o f the sea.

FIQ. 12.

nearly all the cases in which fog occurred when the air temperature was
less than the water temperature, the wind was very light,'so that a kite
ascent would have been impossible.

TAYLOR-THE

FORMATION OF FOG AND MIST

257

The connection betweell wind force and the difference between air
temperature and water temperature during a fog is shown in Table 11.
It mill be seen that the connection is very marked, and would fully
account for the fact that all the fogs explored by means of kites were of
TABLE

I.--YHEQUENCP

OF OCCASIOSS

WHEN A I R

TEIPEILAWRI<
WAS

CREATE&

TIIAB, EQUAL TO, AND LESS THAN WATER TEAlI'EItATURE DURING FOGGY A I D

DURING

CLEARWEATHER.
Nunher of cases in Cases in which air
which air was
temperature differed
greater than sea
from sea temperature
teniperature.
by 1C. or less.

Observations.

141 observations in fog

Percentages

"3

80.2

Percentages

. .

temperature was
less than sea
temperature.

I4

14

9.9

9.9

396

137

72

596

29.6

10.8

665 observations in clear

weather

Cases in which air

the type produced by warm air passing over a cold sea. No other
definite results were obtained regarding the fogs which appear to be
produced in cold air when i t blows over a warmer sea.
TABLE II.-FREQUENCYOF OCCASIOSS W I I E N FOG OCCUI:RED WITH DIFFEREKT
W I X D FOItCES, A N D i\fEAN DIFFERENCE
BL'T\VEEN A I R T E I P E R A T U I l E A I D
\\'ATER

TEMPERATUllE I N EACH CASE.

Wind force.
Beaufort scale.

Number of cases.

3
20

30
1

4
3

46

26

Mean value of air

temperature minus water
temperature.

- 0.5

+0.3
+ 1.0
+ 1.0
+ 1.1
+ 1.8
+ 2.6
+4'I

FOGSON LAND.
We now turn to the last, the most complicated, arid, unfortunately,
the most frequent cause of fog formation, the cooling of the ground by
radiation to the sky at night.
It is a matter of common observation that fogs occur with light
winds or in calms, and that they usually appear when the sky is clear of
clouds. This suggests that they are formed during the night on the spot
a t which they are observed in the morning.
9

FORMATION OF FOG AND blIST

TAYLOR-THE

258

To find out how far this is true, and what wind velocity prevents tlie
formation of fog, I examined all the fogs which formed in the night arid
ware observed in the morning a t Iiew in the years 1900-1905. This
leaves many fogs unaccounted for, becaiise meteorologists make their
observations at about 7 a.m., and many summer fogs have cleared off by
that time. The wind at ICew is registered by self-recording instruments,
and is tabulated every two hours. For convenience I have divided the
wind velocities into four classes. I n the first class are included all the
occasions when the wind velocity was between 0 and 3.3 miles per hour.
T h e second class contains minds from 3 - 3 to 5.5 m.p.h. ; the third, winds
from 5.5 to 9.2, and the fourth, winds from 9.2 to 13% m.p.h. A t
4 p.m. on the afternoons of the seventy nights on which fog ~vasreported
during the period 1900-1908, there were cxarnplcs of winds of each of
the four classes. The actual number of cases in each of the four classes
is shown in the first column of figures in Table 111. At 6 p.m. the
OF WISD OF VARIOUS STRENGTHS
AT 4 P.M., 6 r.x.,
P.M., AND h f I D S I O I I T ON TIlR SEVESTY OCCASlOiiS \VIlEN FOG W A S
REIOIITED A P KEW IN THE NIOIIT D U l l I N O TIKE Y E A R S 1900-1905.

TABLE

III.--FILEQUENClES

8 P.M., 10

tvinci velocity.

P m.

6p.m.

I
I

0-

3.3 m.p.h.

3.3- 5.5

,I

5.5-

I!

9 2

9 2 - 13.6

.
,, .

.
.

8p.m.

1op.m.

Ihlidnight.

24

23

16

evening calm had set in, and i t will be seen on looking a t the second
column that there were only two cases in five years in which a wind of
9.2 miles per hour or more at 6 p.m. was followed by a fog during the
night. At 8 p.m. tho wiiid had dropped still further on the foggy
nights, so that there wore only two cases in the five years when a wind
of more than 5.5 miles per hour was followed by fog during the night.
At 10 p.m. there were still two cases in tlie third class, while there werc
ten cases in the second class.
It will be seen that Table 111. can be used for forecasting, because if
I no fog had been predicted a t Kew every tinie the wind a t 8 p.m. was
greater than 5.5 m.p.h., only two mistakes would have been made in
five years. \Ve might also have predicted no fog every time tlic wind
at 6 p.m. was greater than 9.2 m.p.h., and again we should have been
wrong twice in five years ; but since there were more occasions when
the wind was greater than 5-5 m.p.h. a t 8 p.m. than there were
when i t was greater than 9.2 m.p.h. at 6 p.m., it is evident that we
should have made more correct prophecies by using the wind a t 8 p.m.
than the wind a t 6 p.m. That is to say, the 8 p.m. criterion that the
wind must be less than 5.3 m.p.11. gives rise to fewer false alarms,
when the possibility of fog is predicted and 110 fog follows, than the
6 p.m. criterion that the wind must be less than 9.2 m.p.h.
. Reasoning in the same way, I found, on examining the Kew records,

TAYLOR-THE

FORJIBTION OF FOG AND MIST

259

that the wind at 8 p.m. was a better criterion than the wind a t 10 p.m.
It seems, therefore, that a fog forecaster would do better by observitig
the wind a t 8 p.m. than by observing the wind either before or after
that hour.
After rejecting all nights wheii the mind a t 8 p.m. is greater than
5.5 m.p.h., there remain a large number of nights when fog does not
form. In many of these cases the sky is covered with clouds. The
earth does not cool by radiiition under these conditions, and consequently
the swface air does not cool either, so that a fog is not to be expected.
If, however, all nights when the sky is covered with clouds a t 8 p.m. or
any later time in the night be rejected, there still remain many eases of
calm clear nights during which no fog forms. The question which must
be looked into is, therefore, why fogs sometimes do form, and sometimes
do not, on calm clear nights. How is it that on two nights which
appear exactly similar, fog niay form on one occasion and not on the
other 1
Two possible explanations a t once suggest themselves : either the
ground is so warm that the temperature cannot go down enough to
produce a fog when the air has the normal amount of humidity, or the
air is so dry that a fog would only be produced if the fiill of temperature
during the night were much greater than is usual on calm clear nights.
The first of these is very difficult to deal with, depending as i t does on
such a complicated series of causes. The second, however, seems to be
the more important, which is fortunate, because it is certainly more easy
for a forecaster to take account of the dryness of the air than to take
account of the conditions which govern the temperature of the ground.
I n order to predict whether fog is likely to appear on a calm clear
night, i t is first necessary to predict the amount by which the temperature
is likely to fall in the night, and then one must measure the amount of
water vapour in the air and find out whether the predicted fall in
temperature is great enough to produce a fog.
A t first sight the second of these seems to be a question which could
be solved by purely physical reasoning. The extreme lightness of a
wind during a fog suggests that the occurrence of fog a t any place is due
to the cooling of the ground in the neighbourhood, and not to a mist
which has rolled over from somewhere else. When stagnant air is cooled,
one might expect the proportion of water vapour contained in the air to
remain constant till the fog begins, when i t mould decrease owing to the
fact that some of the vapour had been precipitated as drops.
On examining the Jiew records I found that on foggy nights the
vapour pressure remained constant for a time aiid then began to decrease.
I n Fig. 13 will be seen a curve which represents the mean vapour pressure
on foggy nights in the summer. It will be seen that this curve shows
the characteristic referred to. . Taking this as evidence tliat the vapour
pressure remains constant till the fog forms, I next proceeded to calculate
how much the temperature would have to drop after 8 p.m. before the
a i r became saturated. I then compared the actual ampunts by which the
teniperatures a t Kew had fallen after 8 p.m. on clear calm nights, and
found, to my surprise, that the temperature of the air near the ground
falls below the dew point at 8 p.m. nearly every night when these
conditions hold. It is evident, therefore, that the vapour pressure must

260

TAYLOR-THE

FORMATION OF FOG AND XIIST

decrease on non-foggy nights as wcll :is on foggy nights. I examined

the clear calm nights when fog did not form a t Kew, and foiind, as
before, that the vapour pressure of the air near tho ground remains
VAPOUR PRESSURE AND T E M P E R A T U R E ON,
CLEAR NIGHT'S FOLLOWED BY FOG

FIG.13.

FIG. 14.
constant during the afternoon and evening, and begins to decrease a t
about 10 p.m. In Fig. 14 is shown the curve which represents the mean
vapour pressure on non-foggy nights in summer.
The decrease in vapour pressure is evidently caused by the deposition
of dew on the ground. The water vapour and heat may be tranEmitted

TAYLOR-THE

FORMATION OF FOG AND MIST

261

to the ground either by up and down currents, or by diffusion and

conductivity. I t is evident that if the dryness caused by the deposition
of dew on the ground difuses upwmds a t a greater rate than the coldness
is conducted upwards, then fog is less likely to form than if the reverse
conditions hold. I t is important, thereforc, to form an estimate of the
relative importance of up and down currents and of diffusion and conductivity in transmitting heat and water vapour from the atmosphere to the
ground.
Theoretically the effects of changes in tcmperatdre at the ground
extend iipwarda by conductivity to all heights instantaneously, but above
a c e r t i n height the effects are very small. Suppose, for instance, that
there :we no vertical currents, and that the ground begins to cool at time
t = 0, and that i t cools at a uniform rate, then a t time t the height z, a t
which the temperature has fallen through &th of the fall a t the ground, is1
z = 0.8

J, J Z
!2

Using the commonly accepted values for

the con-

= 0.186 and the

ductivity, p the density, and u tlie specific heat of air, !Pff

formula reduccs to z=O.69 JtT

The height z may fairly be taken to represent the height above which
the effect of the cooling tiad not penetrated a t timc t. Putting t = 1 hour,
4 hours, and 8 hours in the formula, it will be found that z = 1 ft. 44 ins.,
2 ft. 9 ins., and 3 ft. 10 ins. respectively. It is evident, therefore, that
mithout the assistance of vertical currents the cffcct of the cooling of the
ground would only extend, under the action of conductivity alone, to a
height of ahout 4 feet in the course of a night. Since it is known that
the cooling does extend to vcry much greater heights than this, we must
conclude that vertical currents are more important than Conductivity.
The equations of diffusion are exactly the SiLme as those of conductivity
except that the coefficient of diffusion, D, replaces the quantity 5 in the
F@
formula. The measured value2 of D for the diffusion of water vapour into
air is 0.20. It will be noticed that this is very nearly numerically equal
to = 0,186; it is evident, therefore, that the height a t which the change
P@
in vapour pressure is &th of the change a t the ground is practically equal
to Z. That is to say, the cooling and the drying extend upwards very
nearly a t the same rate under the action of conductivity and diffusion.
If D were exactly equal to the effect of diffusion and conductivity on
PO
the distribution of humidity and tempcrature would be the same as that
of a certain distribution of up and down currents in a non-conducting and
non-diffusing atmosphere, because the heat would, in the latter case, be
conveyed upwards by the same cutrents of air a t the same rate as the
matcr vapour.
This last conclusion is a very useful onc, because it shows us that we
can use the diagram of Fig. 6 in reasoning about changes in the vapoiir
l)rc3sure and tomperature of tho s i r near the ground without considering
whether the changes are brought abotit by vertical currents or by
conductivity and diffusion.
1 Taylor, Eddy Motion in the Atmosphere.
(ser. A.), 1915, p. 8.
* See Kaye and Labys Physical Tables.

London, Phil. Trune. R. Soc., 215

262

TAYLOR-THE

FORhIATION OF FOG

AND MIST

I n order to iise the moisture diagram of Fig. 6 to determine how

much below the dew point at 8 p.m. air must be cooled before fog is
produced, it would be necessary to know all about the UP and down
currents which transmit the heat and water vapour to the ground. It i s
impossible, therefore, to apply the diagram to determine whether fog will
or mill not form on any given occasion without making somc further
assumptions in regard to these up and down currents. I have, however,
been able to obtain a t least one theoretical result from the diagram which
is independelit of 'these currents. It will be found tlmt this result is in
agreement with the Kew observations.
Suppose that before tho ground starts to cool by radiation to the sky,
all the air in the lowest layers of the atmosphere (those which are
affected by the cooling) is in a condition represented by a point P on the
moisture diagram-of Fig. 6. Suppose also, in the first place, that tlie
earth were t o cool suddenly to a tetnperature below the dew point and
were to remain a t that temperaturc, deiv would condensc on the ground,
and the air which rose from the surface after depositing its deiv would
be saturated, and at the temperature of the ground. It would, therefore,
be represented by a point A on the saturation curve. All possible
mixtures of air represonted by A and air represented by P are represented'
by points on the straight line joining AP, nnd morcover evcry mixture
of any of the mixtures is also represented by n point on the straight line
AP. The successive states of the air from tlic ground up to tlie
greatest height affected by cooling is therefore represented by all the
points on AP from A up to P. This proposition is true whatever be
the mechaiiism of the vertical currents which transfer the coldness upwards. The question of whether. fog will form undcr the conditions
supposed will therefore depend on whether any of the straight line AP
lies on the upper side of the saturation curve. This is a purely
geometrical question.
From P draw a tangent to the saturation curve touching i t in the
point T. If A is lower down the saturation ciirve than T, then part of
AP lies on the upper side of the saturation curve, while if A lies above
T on the saturation curve the whole of the linc AP lies in the unsaturated region.
I t appears, therefore, that for every condition of the air there is a
certain tempernturc (defined by the point at which the tangent from the
point P which represents this condition to the satmttion curvt) touches
the curve) to which the ground must be cooled, if it is cooled suddenly,
in order that fog may be prodiiced.
The assumption that the cooling takes place instantaneously, and that
the ground then remains at the low temperature while tho coldness
diffuses into the air, is the most unkvourable condition for fog forma t'ion.
If the air is cooled gradually from the temperature represciited by the
abscissa of the point P (Fig. G ) to the temperature represented by the
abscissa of the point A, then the curve which represents the states of the
air from the surface of tho ground up to the region into which the coldness has not penetrated lies necessarily in the triangle PAD, wherc D is
the dew point corresponding with P ;. for the air must consist entirely of
mixtures of air whose states are represented by the straight line PD and
See Appendix.

TAYLOR-THE

FORRIATION OF FOG AND MIST

263

the portion of the saturation curve which lies between A and D. The
actual form of a possible curve is shown as a dotted line in Fig. 6. It
will be seen that as A moves down the saturation curve towards T this
curve will cut the saturation curve before TA becomes a tangent to the
curve. We have, therefore, the following general proposition : If the
temperature of the earths surface falls below a certain level, defined by
the point T, fog will necessarily be produced. If the temperatwe falls
below the dew point, but not so low as that defined by the point T! then
fog may or may not be produced according to the circumstances attending
the process by which the cooling is brought about.
I have verified tliis proposition from the Iievv observations, and also
from some Farnborongh obscrvations. I took every case during an
intervd of five years a t Iiew, when the wind velocity a t 8 p.m. was less
than 8.5 miles per hour. I marked on a moisture diagram the point P
corresponding with the hygrometric state of the air at 8 p.m. and found
the temperature corresponding with the point T where the tangent from
P touched the saturation curve in each case. I found that on nights
when fog appeared the grass minimum thermometer sometimes registered
a lower temperature than this and sometimes a higher. On nights when
no fog appeared, however, tho grass minimum never fell below it.

FOGFORECASTING.
The temperature defined by the point T is interesting theoretically,
but i t is difficult to mitke use of it, in devising a practical scheme for
forecasting fog. For this purpose I have tried to find out a simpler,
more empirical mcthod. It is well known that the temperature falls
more riipidly on warm summer evenings than i t does in the winter when
i t is cold. Suppose that we make observations a t 8 p.m., and that me
know the average fall in temperature during a calm clear night for any
temperature a t 8 p.m. Suppose also that we know how many degrees
below the dew point the air must be cooled before fog forms ; then for
every temperature a t 8 p.m. there will be a certain amount of moisture
(or n certain vapoiir pressure) such that if the air is moister, fog is
likely, whereas if the air contains less moisture, fog is improbable. One
could therefore dram a line on a moisture diagram which has the
following property: If the point which represents the state of the air
at 8 p.m. lies below it, fog is unlikely during the night, while if the
point lies above it, fog is probable.
Now although we do not know the data by means of which me
could construct the line, j e t since we know theoretically that such a line
should exist, we can find i t empirically by marking all the points representing the state of the air a t 8 p.m. on calm clear nights on the diagram,
and drawing a line to divide, as well as possible, the points which represent foggy nights from those which represent non-foggy nights.
As a matter of fact this form of moisture diagram in which the state
of the air is represented by a point giving the vapour pressure and the
tempcriture is not very convenient.
Vapour pressure is not, as a rule, measured directly. The most convenient way of measuring the amount of water vapour in the atmosphere
A diagram wliich shows the vapour pressure and temperature of the air.

264

TAYLOR-THE

FORbIA'HON OF FOG AND MIST

is to measure the difference in temperature between a dry thermometer

and one wliose bulb is kept wet by being covered by wet muslin. \\'hen
the air is dry there is considerable difference in temperature between the
wet and dry bulbs owing to the cooling of the wet bulb by evaporation.
When the air is saturated, the water does not evaporate from the wet
bulb. The wet bulb is therefore not cooled below the temperature of
the dry bulb. The difference between the temperatures of the wet and
dry bulbs therefore measures how far off the air is from saturation.
Tables have been drawn u p showing the vapoiir pressure and depression
of the dew point' for various temperatures of the wet and dry bulb
thermometers.
Using these tables I took every case in which the wind and cloud
conditions a t 8 p.m. at Kew were suitable for the formation of fog, ic.
every ease in which the wind was less than 5.5 m.p.h. and the sky mas
not covered with clouds, and found the depression of the dew point. I
then marked each of these on a diagraiii representing the state of the
air by points giving tho temperature as abscissa and the depression af
the dew point as ordinate. For every night when fog occurred I marked
the point as a round dot, and for every night when no fog was reported
I marked a cross. The diagram is shown in Fig. 15. I thcn drew the
line shown in the figure which divides roughly the dots from the crosses.
It will be seen that there are only 3 dots above the line, while there are
31 crosses. If this diagram had been used on all occasions when the
wind and cloud conditions were favourable for fog, absence of fog owing
to insufficiency of moisture would have been predicted on 34 occasioiis,
and only three mistakes aoiild have been made. On the other hand,
when the point lies below the line, the chances are about equal that a
fog will occur n t Kew and that i t will not.
It appears, therefore, that a diagram of this form should be of some
use in predicting the occurrence of fog if it applies to other places
besides Kew. I find on examining the records of Oxford, Nottingham,
and Potsdam that the line is in each case very close to the Rew line.
As the diagram stands, it is not very convenient to use, because,
after reading the wet and dry bulb thermometers, one lias to look up in
a table to find the depression of the dew point. It can be made more
Convenient by transferring the line to a diagram in which the state of
the air is represented by a point showing the difference between the wet
and dry bulb readings as ordinate and the temperature as abscissa.
A diagram constructed on this principle is shown in Fig. 16. The
full lines give the depression of the dew point. They are merely
taken from the tables and are put on the diagram for convenience. The
middle dotted line shown in Fig. 16 was transferred from Fig. 15 by
finding from the tables the difference between the wet and dry bulb
readings corresponding with points in Fig 15.
The diagram may be used conveniently for predicting whether fog is
likely to appear on an aerodrome or other flat piece of land during the
night. The wet and dry bulb readings are taken at 8 p.m. If the wind
is less than 5.5 m.p.h. and if the sky is not completely covered with
clouds, then a fog is probable if the point on the diagram which repreThe depression of the dew point is the number of degrees through which the air must
be cooled before dew is deposited.

TAYLOR-THE

FORJIATION OF FOG AND MIST

sents the state of the air at S p.m. lies below the line.
above the line, fog is improbahle.

265

If the point lies

**

.I

266

TAYLOR-THE

FORMATION OF FOG AND MIST

If the point lies close below the line, i.e. if there is still a considerable difference between the wet and dry bulb readings, but not enough
to ensurc the absence of fog, fog is probable, but since the air is still
unsaturated fog mill not appear for some time after the readings aro
taken. A difference of more than a degree between the wet and d r y

bulb readings may be taken t o indicate tliat fog will not appear immodiately. A line has been drawn on the diagram which indicates roughly
the difference which is necessary in order that fog may not appear for
about four hours, This line cvidently lies below the other line, and
points which lie between the two indicate a probability that fog mill
form after midnight. It seems possible that thc diagram may be of
some use to flying men in estimating how long an aerodrome is likely to
remain free from fog.

TAYLOR-THE

FORMATION OF FOG AND MIST

287

It may sometimes be convenient to make a forecast at 6 p.m. instead

of at 8 p.ni., though for reasons already given the 8 p.m. forecast is
more likely to be correct. To allow for this contingency a third line has
been drawn on the diagram above the 8 p.m. line. W h e n t h e point
representing the state of t h e air at 6 p.m. lies above this line, fog is
improhable during the night.
The chanco that a prediction made with this diagram will b e right
can be judged by looking at t h e diagram of Fig 15 ; but there is another
cause which limits its usefulness in the winter, t h e freezing of water
which is supplied to keep t h e wet bulb from becoming dry. W h e n this
water freezes, the readings of the wet bulb become unreliable. I n t h e
summer, however, t h e use of the diagram should lend t o a considerable
improvement in forecasting the occurrence of a local radiation fog.

APPENDIX.
Proof that the point representing the iiiirture of two inasses of air of
dilferent Iiumidities mid at different teniperatures lies 011 the straight line
joining the two points which represent the two masses of air on a nioisture
diagram in wlrich tlie ordinates represeiit vapoar pressure and the abscissa
temperature.

Let the voluines of the two nin~sesLc V, and V2,

their absolute teiiiperatures
T, and T?,
their densities
P, and Pa,
and their rapour preseures
P, and pn.
Let the teniperature of the mixture be T,
and its vnpour pressure
P.
Theu if u be the specific heat of air at constant pressure, the lieat lost by
V, is V,pl(T, - T)u and the heat gnined by V, is V9pn(T- T&u, and since no
lieat is lost or gnined by the air as a whole, these are equal.
I t is a property of gases that a t a given pressure their densities are
inversely proportional to their absolute temperatures.
Hence
and since

Hence

Vipl(T1- T)u = V,p&T - T&,

.: (T,- T)= v-(T
2
- T,).
T2

V,+V,=

v v T
(++<)

But the gss V, shrinks in volume to

,>
:

'

(1)

while the gas V, expands to

.
T
2
:
,
'
- Hence the total rolunie of the mixture is
(1) the tots1 volume is V, +V2, that is to E ~ Y , the total volume remaim
unaltered.
Now by the law of partial pressures the vepour pressure in the niixture
due to the vapour originally in V, is

TAYLOR-THE

285

FORbIATION OF FOG AND MIST

Similarly the partinl pressure due to the water vapour origiiially in V, is

Hence
or from (1)

Rot from ( 1 )

Hence
Hence if two niasses of air be represented by points B and C (see Fig. 0)
whose co-ordinates in the moisture diagram are p , T, nnd p? T,, then the
point E which represents the inixture, and has co-ordinates p , T, lies o n
the straight line joining J3 and C.