We have shown the decomposition of 
H records from WMO/IAEA into their
principal components yields factors that are identifiable as northern and
southern hemisphere functions of time and allows the user to combine them
with loading estimates, in appropriate measure, according to their location
on the globe. The application of this approach is more flexible, particularly
for those researchers in areas of the globe far from a 
H sampling
station, than the approach promulgated by Weiss and Roether [1980].
These factors can easily be seen as corresponding to the northern and
southern hemispheres by the distribution of their factor loadings with
respect to latitude (Figs. X and Y). The two factor scores combined account
for 90% of the total, zonally averaged, variance in the 
H data set
after normalization. The error in integrated tritium, between our factor
analysis and actual data, is mostly in the 3 to 6% range (see Fig. X),
although there are some locations with greater error. The utility of this
approach lies in the fashion in which these factors can be used.
By locating a position on the maps of 
and 
the spatial
coefficients (factor loading estimates) are combined with the time series
basis functions 
and 
(factor scores) to
make an estimate of the time history of 
H delivered from the atmosphere
at that location. A comparison of this approach to the individual stations
for which actual measurements exist reveals a statistically better fit than
the approach of Weiss and Roether [1980] for most of the stations.
In the paper we present, in tabular form, the normalized factor scores for
the northern and southern hemispheres by year. We present, as maps, the
coefficients necessary to convert the normalized factor scores into 
H
precipitation time series for an arbitrary global location. We also present
factor loading estimators for actual station locations for error estimation
of our model function. By analyzing monthly data against the factor score
approach we estimate the seasonal changes in 
H delivery. A third table
presents this seasonal analysis, including the amplitude and phase shift of
the maximum 
H delivery. In Fig. X we show the global distribution of
this phase shift consistent with the interaction between continental
re-evaporation and stratospheric supply of 
H [ Ehhalt, 1971].