Hottest summers and coldest winters from 1900

"Summer" is defined meteorologically as June, July, and August. Several make a good summer (e.g. temperature, sunshine, and amount of rain), and not all of these things are too everyone's liking. And of course it can be too hot for most people.


Hottest summers

The hottest summer of the twentieth century was 1976, by a considerable margin. Here are the means (taken as mean CET temperature for the three summer months) for the top thirteen:

1976 17.8

1995 17.4

2003 17.3+

2018 17.3-

2006 17.2

1983 17.1

1947 17.03

1933 17.00

1911 16.96

1975 16.9

1997 16.57

1959 16.57

1955 16.5


Hot summer months

Are "summers not as good as we remember them"? No: they're better. Consider this table of the distribution of the number of hot summer months (with CET equal to or greater than 17.5C) across the decades of the twentieth century:

Here are the hot months (more than 17.5C):

July 1900, July 1901, July 1911, August 1911, July 1921, July 1923, July 1933, August 1933, July 1934, August 1947, July 1955, August 1955, August 1975, July 1976, August 1976, July 1983, August 1984, July 1989, August 1990, July 1994, July 1995, August 1995, August 1997, July 1999, July 2003, August 2003, August 2004, July 2006, July 2013, July 2014, July 2018, July 2019, August 2020.

And here are the very hot months (more than 18.5C):

July 1921, August 1947, August 1975, July 1976, July 1983, July 1995, August 1995, August 1997, July 2006, July 2018.

And the extremely hot months (more than 19.0C):

July 1983 (19.5), August 1995 (19.2), July 2006 (the hottest of all, 19.7), July 2018 (19.1)

So the only really outstanding British summers, with two consecutive hot months, have been 1911, 1933, 1955, 1976, 1995, and 2003. So much for the great British summer! Note also the scarcity and pretty random distribution of these hot months. I have written about our nostalgia for non-existent weather of yore elsewhere (Harley, 2003, "Nice weather for the time of year").

Interestingly, although extremely hot days often occur in early August (without exception until 2019), the extremely hot months are often July.

Before the twentieth century, summers were worse. the only very hot month before 1921 were July 1783, with a mean of 18.8C, and 1852, with a mean of 18.7C from when records began in 1659).

Warm summers have become much more common since 1990, even if some of them have been wet. The worst summers since 1900 are 1922, 1954, and 1972.


Consecutive days over 30C

1976 was clearly exceptional on other measures: in the summer of 1976, somewhere reached the 90s (32C) for 15 consecutive days from 23 June and 7 July (before or since there has been no such spell longer than give days over 90F in a row). There were 3 consecutive days over 35C (26-28th June).

In 1995, in addition to 7 days over 31C, on 32C was reached on 6 consecutive days from 29th July-3rd August .


1983 had 6 days over 32C from 11th-16th July.


1990 had 4 32C+ from 1st-4th August, and like 1976 and 2020 only, had 3consecutive  days over 35C.


1906 had 4 32C+ from 31st August-3rd September.


Heathrow had 16 consecutive days over 30C between 23 June and 8 July 1976, a record.

There were 14 consecutive days over 30C in August 1947, and there had also been 6 in a row 26 May to 3rd June.

In August 2003, from the 5th to the 13th, there were 9 consecutive days where somewhere in the country exceeding 30C, the longest such run since 1976, and 29C was exceeded between the 2nd and 14th - the longest run since August 1997, and 10 consecutive days over 30C in from 3rd - 12th. 32C was exceeded on 5 days.

In 1975, there were 8 consecutive days over 30C from 2nd-9th August which was preceded by 4 days from 28th-31st July and then followed by another 4 from 11th-14th August.

1983 saw 7 consecutive days over 30C from 11-17th July.

The end of June 2018 saw 5 consecutive days over 30C at the end of the month, Porthmadog in Nort Wales doing particularly well.

August 1997 saw at least 5 consecutive days from 8th-12th recording over 30C.


July 2006 saw 30C reached each day between 16th-29th apart from 23rd.


August 1997 saw 5 consecutive days from 8th-12th recording 30C+.


July 2006 saw 30C reached each day between 16th-29th apart from 23rd.



The coldest winters

Which are the coldest winters of the twentieth century? Here are the top ten.

1962-63 -0.3

1946-47 1.1

1939-40 1.5

1916-17 1.5

1978-79 1.6

1928-29 1.7

1941-42 2.2

1940-41 2.6

1981-82 2.6

1984-85 2.7

Are winters not as good as they used to be? Not lately they haven't, but it isn't clear that there is any trend until very recently.

Let's consider the distribution of severe winters, defined by two criteria.

1. Severe months with average temperature less than 2.0C. In the first half of the century, there were 14 severe months, and two occasions (Jan-Feb 1917 and 1929) when there were consecutive severe months. In the second half, there were actually a few more - 17 severe months, with two occasions (1963 and 1979) when there were two consecutive severe months, and only one occasion with three consecutive severe months (1916-1917). This difference however will be statistically insignificant.

2. Very severe months with average temperature beneath freezing. In the first half of the century, there were two very severe months (January 1940 and February 1947. In the second half, there were five (February 1956, January 1963, February 1963, January 1979, and February 1986 - and December 1981 came very close).

I think two things are happening. First, people's memories are distorted by the exceptionally severe winters of 1947 and 1962-63, in particular. Second, I think there is a recency effect where there haven't been many severe winter episodes in the last few years. Consider this table of the distribution of the number of severe winter months across the decades of the twentieth century:

Older people might be affected by the bump in the middle of the century (particularly the war years, which were exceptional) and compare that with the 90s. There is no overall trend, however: note the scarcity of severe months in the 0s and thirties. It is also possible that central heating means that we are less sensitive to cold weather when it happens.

Here are the severe months (mean less than 2.0C):

Feb 1902; Dec 1916, Jan 1917, Feb 1917, Feb 1919, Jan 1929, Feb 1929, Dec 1933, Jan 1940, Jan 1941, Jan 1942, Feb 1942, Jan 1945, Feb 1947, Dec 1950, Feb 1955, Feb 1956, Jan 1959, Dec 1962, Jan 1963, Feb 1963, Feb 1968, Feb 1969, Jan 1979, Feb 1979, Dec 1981, Feb 1983, Jan 1985, Feb 1986, Jan 1987, Feb 1991, Jan 2010, Dec 2010.

Here again are the very severe months (mean less than 0C):

January 1940, February 1947, February 1956, January 1963, February 1963, January 1979, February 1986, December 2010.

It is true that although we have some recent hot months, until 2010 you have to go back to February 1991 for a "severe" cold month, and 1986 for a "very severe". This difference is a clear suggestion of change.

Before the twentieth centuries, winters were better (if you like then cold and snowy). Remember the Little Ice Age? In the seventeenth and eighteenth centuries, winter knew how to be winter. Take the coldest winter of them all, the "Lorna Doone" winter of 1683-84, with a CET of -1.2. Or 1739-40, with a CET of -0.4. Now that is chilly.

Cold winters have become less common since the end of the 80s.

We have at last had a cold winter, with 2009-2010 giving an average of 2.4, which would have made it 8th in the above list, and the coldest since 1978-79. January 2010 counts as a "severe" month (at 1.4C).

And of course  December 2010 at last gave us a very severe month, with a CET mean of -0.7, the first such month with a negative CET average since February 1986.

The coldest of all months in recent times remains January 1963 (-2.1C), although January 1814 (-2.9) and some months in the Little Ice age were almost certainly colder, with January 1795 around -3.1C.