1298 
nates the particular signal to be located, choosing 
those relatively distinct from others in the typically 
rapid sequence; indication of simultaneity and relia- 
bility is obtained in the plotting on gnomonic maps. 
Observations are generally made at two- or three-hourly 
intervals for periods of about half an hour. Base lines 
connecting stations may range in length from a few 
hundred miles to over a thousand. 
A second direction-finding system, developed by Lut- 
kin [20], also employs a pair of loops at right angles 
to one another, so arranged that a record of the in- 
coming sferics signal is made only when the plane of 
one of the loops is directed toward the source. The pair 
of loops rotates at perhaps one revolution per minute, 
and the system provides a record on a drum turning 
with the loops. The receptive sector may be reduced 
to one degree of azimuth. Records obtained in this 
manner show direction of signal arrival as a function 
of time; a rough indication of the proximity of the 
source can be gained from the relative size and density 
of the marks made by the recording pen. The narrow 
sector system has been employed by Lugeon [16] and 
Bureau [8], among others. It provides very simply a 
record upon which statistics might be based, and rough 
conclusions may be drawn regarding the location of 
sources of sferics from observations made at several 
stations, or from the appearance and density of the 
record at one station. This system has found use in 
Switzerland, France, Australia, and other countries. 
Other systems measure and record, as functions of 
time, the frequency of occurrence of sferics having an 
intensity greater than a present level (Bureau’s radio- 
cimemograph and Lugeon’s atmoradiograph) [8], and 
the field strength of sferics (Lugeon and Nobile’s radio- 
maximograph) [18]. Similar systems provide power com- 
panies with information on the proximity and course 
of intense local storms [29]. Rough determination of 
distance of sferics sources from single stations em- 
ploying these and the narrow sector systems is made 
possible by astronomical methods involving the ozono- 
sphere or ionosphere [14, 18]. Observation at a single 
station of the wave form of sferics indicating multiple 
reflections from the ionosphere also permits rough de- 
termination of distance [15, 23, 33]. Approximations of 
distance have also been based upon the assumption 
that the most intense discharges in all but the weakest 
storms have about the same average power [18]. 
Accuracy of determination of bearings of sferics varies 
diurnally because of changing polarization of the ar- 
riving radio wave. These errors are largest at night, 
and especially at sunrise and sunset [22, 23]; neverthe- 
less, useful data may be obtained even then, although 
the efficiency of observation is reduced. Polarization 
errors are greatest in an intermediate range between a 
few tens of miles and some 1500 miles from the re- 
ceiver, and are smaller at the very low than at the 
medium and high radio frequencies. Most other sources 
of error, due to nonhomogeneity of the site with respect 
to azimuth, and to instrumental and operational errors, 
can be made negligible. Use of antennas other than 
loops to decrease polarization errors is being investi- 
RADIOMETEOROLOGY 
gated [1]. Regarding propagation, range, and attenua- 
tion of low-frequency radio waves, see, for example, 
[4, 5, 7, 10, 21]. In triangulation procedures with data 
obtained from direction finders, the errors of closure 
of polygons formed by laying out the several reported 
azimuths, together with observers’ estimates of con- 
fidence in the value and synchronization of each azi- 
muth, have been utilized empirically to assign estimates 
of accuracy to sferics fixes. Error charts based upon 
the geometry of a network with respect to any source 
of sferics have been constructed. These show which 
pairs of azimuths yield the smallest maximum error, 
under the assumption that there is a constant error 
(generally taken as + 2°) in azimuth determination 
at each station. In network operation, the task of 
designating which sferics shall be observed is often 
rotated among the stations in the network, in order 
to avoid favoring a circle of detection about only one 
station, and to reduce the blanking effects of local 
storms. Probable errors are assigned for each fix based 
upon location with respect to the observing network, 
estimates of instrumental error, size and shape of tri- 
angle or polygon of intersection of bearings, and con- 
sistency. Significant contributions regarding selection 
of an optimum point for a fix, given several radio 
direction-finder bearings, have been made by Ross [28], 
Stansfield [34], and Barfield [3]. 
Of what value are sferics to meteorologists? Like 
radar, sferics instruments provide the equivalent of 
an extremely dense network of observing stations within 
the working range; the practical need for such high- 
density coverage depends upon the significance of the 
weather element(s) detected, the size of the area 
covered, and the density of the regular weather ob- 
servational network over the area. Patterns of activity 
are more readily perceived than by use of point ob- 
servations. In the present case, the forecaster insures 
that his map analysis, especially when based upon 
widely scattered observations, agrees with the fact of 
occurrence of vigorous convection at certain places in 
the pattern. Synoptic maps were revised on this basis 
as early as 1924 [4]. Petterssen and Berson [26] in a 
recent study of European and North Atlantic sferics 
reports and winter synoptic maps found concentrations 
of sferics sources at the apex of warm sectors of both 
nascent and occluding cyclones, in the former case 
even before the appearance of closed isobars at the 
surface. Maxima were also found on the cold front 
400-500 miles from the apex, and secondary maxima 
300-500 miles behind the cold front. 
Study of the results of observations of a network of 
direction finders in South Africa led Schonland and 
his collaborators [32] to conclude that 70 per cent of 
sferics fixes could be verified by local observations of 
thunderstorms and precipitation, though perhaps more 
were real, but that the utility of sferics reports in 
forecasting tends to be marginal. 
The performance of a United States direction-finding 
network consisting of stations in New Jersey, Florida, 
Newfoundland, and Bermuda has been studied [38] 
by comparison of sferics fixes with reports of thunder- 
