20 
g) Synthesis 
Synthesis is effected by equations (1) and (2), respectively. The question 
remains open, for which prediction and retrodiction time periods the coeffi 
cients retain their validity. After all, these coefficients do by no means represent 
“constants”. Tides are natural events and are subject to steady albeit minimal 
change. In particular with regard to gauge time series of tidal rivers it is recom 
mended to conduct synthesis on the basis of analyses of the immediately pre 
ceding 19 years, in as far as these are already available. If one was to calculate 
post hoc syntheses for a longer period of observations, one can determine 
coefficients spanning 19 years and use these to respectively generate only the 
central year. The syntheses for the first and last nine years then are calculated 
by means of an unchanged set of coefficients. 
When generating tide tables for tidal rivers, especially in the years following 
fairway adjustments, it has proven useful to generate particularly constant 
members (Table 2), e.g. from previous years and to disregard in synthesis (1) 
and (2) the long-periodic proportions with j { and j 2 (Table 5). Such optimisation 
of the here presented method leaves room for further development work.