This article was previously published asManley (1947).

Some time must always elapse before a full account can be compiled of a ‘historic’ spell of weather, during which other events commonly occur to catch our attention. February 1947 was, however, so exceptionally memorable that a further review of some of its features is well justified. As a result of its exceptional cold the mean temperature for the three months, December–February, ranks the last winter about the tenth in order of severity in the past 200 years.

With regard to temperature, monthly averages for long periods have been compiled for several British districts. The Stevenson screen has, however, been in use since about 1875. Before that date various methods of exposure of thermometers were in use, none wholly satisfactory. Hence it is by no means easy to bring old records of temperature to modern standards. In recent years, however, a number of comparisons between older and newer methods have become available, from which allowances can be derived. Moreover, before 1850 fixed-hour observations were in vogue, twice or thrice daily; and if the hours were well chosen, fairly satisfactory values of monthly means can be obtained. Thirdly, in certain instances, older types of screen have been kept in use; for example the north-wall screen at Kew, 17 feet above the lawns, gives minima appreciably higher than those in a Stevenson screen with normal exposure. But such records provide us with ‘bridges’ across the periods during which more modern screens were being brought into use.

A further difficulty arises from the lack of homogeneity with regard to the site of stations; it is well known how greatly the extremes may differ, especially in winter, between stations quite close together. It is not fair to assume that, because a station at Liverpool recorded 15°F in January 1780, while Bidston recorded 15°F in January 1940, Walton (Liverpool) recorded 10°F in January 1795 and Birkenhead 11°F in January 1838, that closely comparable degrees of severity existed. The first record was kept on a roof by the docks; the second in a modern screen on a hill-top in the Wirral; the third was on the north wall of a house; the fourth, partly sheltered, on a garden wall close to the Mersey. Many allowances for location and manner of exposure must be made.

Reduced values, even of averages, are therefore not often to be found. The dates at which some long British tables begin may be mentioned. First, stations at which the site at least has been virtually unaltered include Oxford (1815), Greenwich (1841), Durham (1847) and Stonyhurst (1848). Second, other tables have been compiled with the aid of several overlapping records; these include ‘Lancashire’ (1753), ‘London’ (1763), ‘Edinburgh’ (1764) and ‘Kew’ (1783). The further we go back before 1800 the more caution is necessary with regard to the acceptance of records. Abroad, easily the most remarkable table of monthly means is that recently published by Dr. Labrijn for Holland, beginning in 1706. For Sweden, tables are available from 1739.

To glance for a moment at the other climatological elements: rainfall records back to 1860 are abundant. Beyond that, ten or more English stations, with some in Scotland, are available for each year from 1800. Still earlier, overlapping records back to 1727 have been used by Dr. J. Glasspoole to give us a survey of the general rainfall for England. With regard to the frequency of snow or sleet, there are many gaps in our records. It is of interest, however, to note that from 1728 to 1748 the frequency of snow in S. Devon was exactly equal to the present-day average: from 1782 to 1820, however, snow seems to have fallen in N. Scotland considerably more often than in the past 30 years. The occurrence of snow-cover has only been tabulated since 1912. Wind directions at a few places carry us back well before 1750. Barometric readings abound, but very many of the early records are almost useless as no temperature corrections were made. We can, however, derive a very good idea of the day-to-day fluctuations; and any meteorologist can put together the pressure changes, clouds, winds and temperatures and build up a very good picture of the air masses involved in the weather 150 or more years ago, although the accounts of those days are extremely scattered in out-of-the-way journals.

The following table of approximate monthly means for representative stations covers the eight outstandingly severe months since 1790 [Table 1]. It is clear at once that the departures from normal are of the same order throughout. So long as the North Sea does not freeze over, the monthly means for lowland stations cannot depart far below the freezing point, even at stations well inland; perhaps 24°F represents the limit. It is not unusual to find that in a very cold month, which is at the same time dull, the Edinburgh mean is slightly higher than that at London, the fundamental reason being the greater breadth of the North Sea. For similar reasons the extreme south-west of England is liable to receive a good deal of air which has travelled with a fairly long ‘fetch’ down the Channel, or has come round a depression in the region of the Bay of Biscay. Accordingly, we find that in a very cold month the mean temperature at Plymouth is roughly 4–7°F higher than in the Midlands. If the values in the table are plotted they show many similarities of pattern.

Table 1. From Edinburgh southward: approximate comparison of the coldest months since 1790 (°F).
 Jan. 1795Jan. 1814Jan. 1838Feb. 1855Jan. 1881Dec. 1890Feb. 1895Jan. 1940Feb. 1947
  • Figures in brackets are not fully standardized. Other very severe months ranking in some districts with the above include December 1788, 1796 and 1878; January 1776, 1780, 1784, 1820 and 1879.

  • a

    This mean was recorded at Teignmouth.

‘Edinburgh’ 29.9 26.5 30.6 30.6 29.1 35.3 31.2(31)(31)
Durham — — — 28.1 28.6 33.4 29.4 29.1 30.1
‘Lancashire’ 26.0 25.6 29.9 28.9 28.3 31.8 29.0 29.8 29.9
Wrexham — — — — — — —(29.0) 27.9
Nottingham —(25.6)(29.3) — —(30.8)(28.6)(29.5)(28.9)
Oxford — — 28.6 29.0 30.3 29.2 28.5(29.8)(28.1)
Cambridge — — — — 29.5 28.0 28.8 29.3 28.5
‘Kew’27.4 29.1 29.6 29.6 31.8 30.0 29.4 31.7 30.3
‘Greenwich’26.3 27.0 28.6 29.2 31.7(29.8) 29.0 30.5 29.3
‘Exeter’ —(30.4) — — 30.6 32.0 29.3 32.3 30.9
Plymouth —(31.8) —(33.4)a(34.3) 36.5 32.4a 37.2 32.9
Falmouth — —(36.4)(36.3) 37.7 38.8 34.7 40.3 35.5
Penzance —(35)(39.0) — — — 35.5 41.9 36.3
‘Utrecht’22.1 26.1 20.1 21.7 27.1 23.4 25.9 22.1 —
‘Paris’20.7 31.6 23.7 32.2(29.7)(26.1)(24.0) — —
Some recorded extremesKendal <8Basingstoke<–3Chiswick–4Chiswick0Lancashire–7Rugby1Buxton–11Rhayader–10Woburn–5
  Kensington<3Walton on-Thames–14 Kelso–16 Braemar–17Ambleside–6Hereford–2
Prevailing characterFog, fair to snowyDull, very snowyDry, little snowVery frosty, some snowFrosty, snowy in southVery dull, dry, quietFine, frostyFair, frostyDull, very snowy

It is broadly true that two kinds of severely cold weather occur with high pressure over the Continent. If the pressure gradient for easterly winds is not so marked, or if the winds mainly reach us travelling by the ‘short sea route’, there may be a fair proportion of days and nights with clear sky, as in January 1940, February 1929 or February 1895. If, on the other hand, the ‘high’ is more persistent over Scandinavia and there is a good deal of turbulence in air with a long fetch over the Baltic and North Sea, cloud is more extensive (e.g. February 1942); and if depressions repeatedly approach our south--western shores, the consequence is the dull, cold and snowy type of weather experienced last February. In the south-west and south heavy snowfalls occur (in the ‘classical’ style of January 1881) on the north side of stationary or very slow-moving fronts. The slow forward and then retro-grade motion of such fronts produced the classical ice-storm of January 1940 in the Midlands, and nearly similar phenomena in the southern counties (Sussex and Devon notably) in February to March 1947. Far back in 1740 an ice-storm in N. France accompanied the exceedingly severe weather of January in England.

Meanwhile further north instability is always likely to develop in the cold air as it crosses the broader North Sea. Hence, although N.E. England and E. Scotland ten-ded, on many days in the winter of 1947, to be slightly milder than the south, there were heavy and persistent snow showers, and the accumulation, accompanied by drifting, especially on all easterly-facing uplands, was remarkable (e.g. Teesdale, N. Norfolk, Huddersfield, Denbighshire and West Bedfordshire at various times during the month). On other occasions frontal development between the milder air over the Irish Sea and the persistently cold land air gave the phenomenal falls in early March in North Wales, the Lake District and Galloway. Sunshine was mainly to be found in Skye and the Outer Hebrides, accom-panied by exceptional drought in the lee of the mountains; at several of the wettest West Highland localities no measurable precipitation was recorded.

The combination of persistent cold air and cloud produced an extremely rare result in our climate; over a very large area, even on low ground, snow lay throughout the month. At Cambridge, for example, it lay for 28 days, whereas the record for any year since 1912 is 29 days. It was, especially in the south and Midlands, a month with few clearances, although over the deep fresh snow-cover inland minima around zero Fahrenheit occurred about February 24–25. More remarkable still was the minimum of –6°F at Houghall, outside Durham, on March 4; here the combination of deep snow- cover, clear sky and the severe frost-hollow formed by the Wear valley gave what, so far, appears to be the lowest March temperature ever recorded anywhere in England. Peebles and Braemar also recorded –6°F.

Further, at many places inland the mean daily maximum for February was below 32°F and the extreme maximum below 40°F. For such a com-bination of low maxima, with very widespread and deep snowfalls, not only in the north-east but also in the extreme south-west, it appears that we have to go back to January 1814. For in other cold months there has usually been much less snow with more clear nights, as in February 1895 and January 1838, 1881 and 1940. December 1890 was much more anticyclonic, quiet and dull, and considerably milder to the northward. February 1855 was moderately snowy, but considerably less cloudy, hence the daytime maxima tended to be higher, and towards the end maxima near 50°F were widespread.

January 1814 seems to have resembled the February of 1947 quite closely: snowfalls were deep and very extensive, although ‘in vicinity of London, not deep – nowhere more than about a foot, in and about Chelsea’. In both months there was very little thaw. In January 1814 very severe frost about 8th–10th and 20th–22nd, combined with the existence of Old London Bridge and the fact that the river was not so well embanked, caused the Thames to become sufficiently frozen to be passable. Quoting a contemporary record, ‘the tide kept the Thames open to Battersea for a long time; above, the river appeared quite frozen over about January 20. The thaw of January 26 (max. 36°F) brought down so much floating ice that the whole river between Blackfriars and London Bridge was filled and when the succeeding frost came on it bound this together so that the river was passable almost anywhere between London Bridge and Westminster’. The east wind continued January 3–26 with little sun. From December 26 to January 2 the fog was exceptionally thick and persistent in the city; it was also observed at Exeter. We can fairly deduce a ‘high’ moving slowly towards the Continent, in a manner not unlike that of the third week of January 1947 or the second week of January 1940. With regard to the earlier dull, snowy period of late January to early February 1795, it is tempting to recognize in the appalling Nottingham and Cam-bridge floods about February 9, associated with very heavy snow further north, a slow-moving warm front, remarkably like the more reluctant affair of March 4–6 1947. Moreover, in 1795 the ice on the Nen was blown up by a courageous volunteer to relieve the danger.

For January 1814 we have the following data from Nottingham (thermo-meter 15 feet above ground, N. wall): mean max. 30.5°F, mean min. 19.0°F; extremes 39°F and 3°F; ‘adopted mean’ 25.6°F. For Derby, exposure uncertain, were recorded 30.7°F, 22.9°F, 37°F and 6°F respectively. For Manchester (Bridge St.) in a sheltered city exposure, mean 27.7°F; extremes 41°F and 10°F. Cary, in the Strand, recorded a mean of 28.4°F. For Carlisle when ‘the most intense frost began on January 3 that probably ever occurred’, and where about six inches of snow ‘very much drifted’ lay for most of the month, the mean (from fixed-hour observations) was 24.5°F. Derwentwater was frozen for six weeks. At Plymouth 11 inches of rain and snow were measured, while in Herefordshire ‘they walked over fences’. The meteorological chart can well be imagined, for on the same day was reported ‘snow and sleet all day’ in London, followed by a strong north-easter on the 20th with renewed frost. This also resembles to a remarkable extent our more recent episode. We may judge, too, from the relative mildness of Paris and Utrecht, that a number of depressions must have approached on tracks much like those of February 1947. At the same time, the Gordon Castle and Edinburgh means (27.0°F, 26.5°F) with those of Carlisle and Perth suggest that January 1814 was somewhat less cloudy in the north than last February, and therefore rather colder. Few of these figures are, however, strictly standardized, and with regard to the accompanying comparative table, it must be remembered that the early mean values (quoted in brackets) are more likely to err on the low side than otherwise.

The freezing of Windermere from end to end is sufficiently rare to offer a very good index of a severe month. In February 1947 the lake was partly frozen, quite thickly in the bays. Yet many spells of weather marked by more extreme minima, and more extensive ice on rivers and lakes, have occurred in the past, e.g. in 1895. But it can readily be seen that the chance of the combination of deep snow-cover with clear skies for more than a few days is very rare in our climate. Therefore a winter month with a mean of the order of 25°F in January, or 27°F in February, probably represents the lowest we can expect unless the North Sea becomes extensively ice-covered.

Many readers who know the Admiralty Pilot will comment that no mention in that tome has hitherto been made of floating ice in the North Sea off the British coasts. The occurrence of drifting ice last February was, therefore, very remarkable. We may indeed observe in these events, and in the per-sistent sluggishness with which fronts moved northward, a hint of the conditions which undoubtedly prevailed during the Pleistocene ice-age. For very many years we shall be able to look back at 1947 and try to seek the fundamental causes of that persistent tendency for high pressure over the greater part of the Arctic, resulting on some days in an almost unprecedented easterly current from Ireland to Newfoundland, which will baffle many old-fashioned believers in ‘south-west anti-trades’.

Although February 1855 was in some places colder, on the ground of low daytime maxima and extreme behaviour, it seems fair to regard February 1947 as the most generally exceptional winter month since January 1814, considering England as a whole. For Scotland this is less certain, in-asmuch as the extreme severity of March 1947 in that kingdom will long be remembered. North of the Highland line it seems likely that it was the coldest March since 1785, although some allowance must be made for the difficulties of equating early temperature data with those of the present day.

And what of the forthcoming winter? Searching the older records in the hope of stumbling across a pattern of succession of events is a pastime in which everyone indulges from time to time. The rarity with which really severe Februaries occur will tempt many to assume that we shall not have another for some time. Assumptions on these lines are, however, very risky. From 1782 to 1786 there were five successive Marches of extraordinary severity. In the 40 years 1898–1939 only two winters might be termed cold by nineteenth- century standards; yet from 1940 to 1947 four have occurred. Following severe spells in January to February, 1795 and 1895 gave, in the main, a dry warm May; 1814 and 1855, a dry cold May; May 1947 we might call warm and on the whole wet except in the south-east – it resembled that of 1917.

Given these contradictions, and many others that soon appear, many would-be seasonal forecasters who search the older records will be tempted to agree with the words of the negro spiritual – ‘Dere's no hiding-place down dere’. Yet in the succession of events following an extreme season there cannot fail to be much interest; and the climatological enthusiast a century hence will surely have a clearer picture of the curious vicissitudes to which we are subject.


  1. Top of page
  2. Reference
  • Manley G. 1947. Looking back at last winter (a) February 1947: its place in meteorological history. Weather 2: 267272.
  1. 1

    Editor's note: for those unfamiliar with temperatures in degrees Fahrenheit, the following equivalences (rounded to the nearest degree) may be helpful:

    –17°F = –27°C; –11°F = –24°C; –6°F = –21°C; 0°F = –18°C; 10°F = –12°C; 21°F = –6°C etc.