At the Met Office we maintain the UK’s climate records and these are used to monitor our climate at a national and regional level. To ensure consistency, weather records are only used from weather observation sites with calibrated, standard instruments and carefully monitored exposure. Although some records have been broken by non-standard stations, these are not accepted as official records for this reason. Also, on rare occasions an extreme temperature is recorded at a weather station which is so at odds with readings from stations nearby that it will be discounted as unreliable.
How far do records go back?
Historical observations of our weather are stored by the Met Office, with individual station observations in our digital database going back to 1853, however there are historical paper records for earlier dates held in the National Meteorological Library and Archive.
When are temperatures officially recognised as a new record?
The historical observations are used for monitoring our climate and maintaining a set of official climate records for the UK. They are also used to highlight interesting facts about our weather and place current weather into historical context. But these arbitrary calendar dates, such as the recent August Bank Holiday, Christmas or Wimbledon, do not give a complete picture of seasonal or monthly extremes.
Each day real-time data is subject to preliminary quality control before it is released, such as cross checking against nearby stations. A record will not become official until thorough quality control has taken place on the data, which may take several months.
They may even be reviewed again at a later date to ensure they remain valid.
Historical records also undergo continual review. With millions of data points it is inevitable that the database contains very small proportion of questionable data.
Was it the warmest August bank holiday on record?
On Monday 28th August a maximum temperature of 28.2 Celsius was recorded at Holbeach. The highest temperature for the late August bank holiday in the digital archive was 28.3 Celsius at March (Cambridgeshire) on 27th August 1990. However this particular reading is considerably higher than all surrounding sites and therefore considered to be suspect. Therefore the August bank holiday of 2017 would be considered as the hottest bank holiday on record beating the value of 27.2 Celsius recorded at East Bergholt (Suffolk) in 1984.
It should be noted that this is a statistical quirk of the August bank holiday never having recorded a notable temperature extreme, and does not reflect a new national temperature record for the UK. We have seen higher temperatures later in the year than this. For example a brief heatwave in September 2016 saw temperatures reach 34.4 Celsius on 13th, and the highest temperature recorded in October for the UK was 29.9 Celsius on 1st October 2011.
For full weather extremes listings and historical records go the Met Office Climate section
Official blog of the Met Office news team 
‘Each day real-time data is subject to preliminary quality control …’
Oh yeah? Where’s the non-in-house, genuine review?
In moderation, of course.
“To ensure consistency, weather records are only used from weather observation sites with calibrated, standard instruments and carefully monitored exposure.”
I take it that includes stations like Heathrow Airport?
http://wp.me/p7wiux-2ib
Where were the other suspect Cambridgeshire stations located and how much higher was the temperature they recorded?
Tonyb
The highest temperature for the late August bank holiday in the digital archive was 28.3 Celsius at March (Cambridgeshire) on 27th August 1990. However this particular reading is considerably higher than all surrounding sites and therefore considered to be suspect.
What, like Heathrow? When you declared 36.7C as the July record there in 2015, it was more than a degree higher than anywhere else in the country.
And then there’s the all-time record at Faversham in 2003 of 38.5C, which was 0.4C higher than anywhere else, even Kew which is heavily affected by UHI.
Even Philip Eden is adamant that the Faversham record is fake, stating that “it is now widely discounted”
But good enough for Met Office purposes evidently!