50 
Probable limit 
of error. 
Earlier records. 
Exceptional 
records. 
Test of 
reliability. 
The ten years’ result appears to he really nearer the truth 
than the 20 years’. An additional ten years made the needful 
correction, so that the results for 30, 40, and 50 years are 
practically identical. But the mean for the 23 years 1861—80 
(48’45), would err even more in excess, than 1841—60 were 
deficient. 
On the assumption that the 50 years’ mean differs only I as 
much as any 10 years’ mean from the real value, the result 
47’72'^ is within 0'21° of the real value. 
The references to earlier records in the Eeports enable us to 
deduce a mean for 1831—1890 of 47'69. The inclusion of 
Mr. Grray’s values (used by Professor Phillips) from 1800—1824 
raise this to 47'83 As, however, these last can hardly be 
considered of equal reliability, we may accept 47'7®,. with a 
probable error considerably less than . 5 °, as the mean of the max. 
and min. thermometers. This, by the correction already 
applied, gives as the mean annual temperature of York 47‘3° pins 
01 ’ ’ 2 . 
In conclusion we may note the following exceptional 
records:— 
> 
The coldest month, February, 1855, had a corrected mean 
temperature of 25'5. In January, 1881, it was 28'P. 
The hottest montlis, similarly corrected, are July, 1852 (64’6^), 
July, 1847 (64 3'';, June, 1846 (63'8°), and August, 1870 
(63-7°). 
In 1856, the thermometer fell to 6 ° on December 4th ; but 
rose to 60" on the 7th. 
In 1868, 80", or above, was reached on two days in June, 6 
in July, 5 in August ( 86 ° on the 6 th), and 6 in September, or 
19 days in all. 
K/ 
Ten years later it was above 80° twice in June, five times in 
July. On June 6 th, the (corrected) mean was 73'4°; on 
December 14th it was 19 9°. 
In 1883, the opening days of May were colder than those of 
January. 
Finally, a comparison of the separate means of max. and min. 
for the five decades may be used as some clue to the reliability 
of the two thermometers. 
♦ 
