th SatialSpanar nabted, - Se Ata atts aise Heat i Be oh at nok ‘teaitlin depth of 3-6 inches. et alice i 5 4 . digit ia ebeoshesfle. PRR ees It 1s difficult to estimate what the accuracy of these observations was, . but assuming that the instrument was correct within 0.2° C., that the observer's readings were & bo correct within 0.2° C., that neither of these possible sources of error was consistently biased and that both errors were distributed normally then 9 values out of 10 are correct within 03° BF Fe and all are correct within 0.7* F. TABLE 8. KEITH: THREE-HOURLY OBSERVATIONS. See Notes for Table 6. : Experienced observers made the cloud and other observations during the following intervalsBF (times are inclusive): 1200 Aug 18 -- 0900 Aug 19; 1200 Aug 21 — 0900 Aug 23; 1200 Aug 26 =-- 0900 Aug 27; TABLE 9 1200 Aug 28 —~ 0900 Aug 29; KEITH: 1200 Aug 30 -- 0900 Aug 31. 1 HOURLY RELATIVE HUMIDITIES. This Note applies also to Tables 22 and 25. The three-hourly values (0300, 0600, etce) are based on direct dry-bulb and wet~bulb readings (Table 8). rsf The remaining values are taken from hygrothermograph charts, with ad just- 7 ments in absolute trace readings being made to fit the three-hourly values. The three-hourly 3 values are al] correct within 6% and 9 out of 10 are correct within 4% (see Notes, Table 4). For intermediate hourly values, these errors increase to 8% and 5%. ‘ Further, at values in the | 80s there is a small bias -- about 1% -- in the direction of giving values that are too low; ! while in the 90s there is similar bias of about 22. Since the hygrothermograph was checked regularly (usually daily and at least every other ‘ day) times are correct within 15 minutes. TABLE 10. MACK: DAILY OBSERVATIONS. This Note applies also to Table 26. rig we OF ee Og sm eg we SS a The Notes for Table 6 apply for all items except RR, DDFF, Sea, and Remarks. Cloud, wind,| sea, and other observations were made by experienced observers on all dates except August 26th | through 29th. RR. j Unavoidably, the raingage was not well exposed (see General Notes and Figure 7). Therefore readings may be in error by as much as 20%, with values probably tending to be too low when the wind at time of rainfall was between NNW and NNE and too high when it was between|