518 
average ly or A» as a function of latitude. They are ob- 
tained by forming means of J, or d» for groups of sta- 
tions within fairly small ranges of latitude. For several 
stations the black dot is ‘‘off” the mean curve by more 
than the length of its uncertainty line. Some such de- 
partures are to be expected from the laws of chance, 
but some certainly indicate that L, does not depend on 
te) 20 40 60 80 
MICROBARS (a) 
COIMBRA 
LISBON 
SAN FERNANDO 
MICROBARS 
(f) 
Oo 10 20 
MIGROBARS 
BATAVIA (d) 
(eee Te So: ae ee ae o 
(0) 20 40 60 80 
MICROBARS 
(c) 
DYNAMICS OF THE ATMOSPHERE 
plitude is on the whole greatest in the J group of 
months. 
Both these features of Lp are illustrated in Fig. 7, for 
the region of North America. Besides the annual mean 
dial arrow for several United States and Canadian sta- 
tions, Fig. 7 gives the forward halves of the dial arrows 
for the J, HE, D groups of months. Not every station 
-- - 
(0) 20 40 60 80 
MICROBARS (b) 
HONG KONG 
oa 25° 6G GO 
MICROBARS (e) 
Fig. 10—Harmonie dials (with probable error circles) indicating the annual change of the lunar semidiurnal air tide in baro- 
metric pressure. (a) Annual (Y) and J, #, D four-monthly determinations for Taihoku, Formosa (1897-1932). Also sets of twelve 
monthly mean dial points for (6) Taihoku (1897-1932), (c) Batavia (1866-1895), (d, inset) mean of Potsdam (1893-1922) and Ham- 
burg (1884-1920), (e) Hong Kong (1885-1912), and (f, in centre) the mean of Coimbra, Lisbon, and San Fernando. 
latitude only, as was pointed out in the preceding 
section. 
Similar mean curves of J. and » as functions of lati- 
tude are given in Fig. 9 for the annual mean of Le, and 
for the group means for the D group of months (as in 
Fig. 8), the J (June solstitial) group May to August, 
and the # (equinoctial) group March, April, Septem- 
ber, October. The individual values of J; and 2 with 
their uncertainty lines are not shown. 
The outstanding feature of the annual change is the 
increased lag of high tide, by about an hour, in the D 
group as compared with the J and & groups. The am- 
shows the two features just described, but this may be 
partly due to the accidental errors affecting all these 
determinations—errors about +/3 times as great as for 
the annual means. 
Figure 10a shows this annual change of Li» in another 
way for a single station, Taihoku, Formosa [43]. It 
shows the dial points and their probable error circles 
for the annual mean L». (marked Y for year) and for 
the three groups of months J, H, D. / 
The annual change is, however, not fully shown by 
such means for four-monthly groups. It is better, where 
adequate data are available, to determine L» for each 
