Published online by Cambridge University Press: 11 May 2009
The evidence for variations in winter phosphate concentration as a component of the ‘Russell Cycle’ has been re-examined. The change from high phosphate concentrations prior to 1930 is the largest apparent variation in concentration through the period 1923–1987. However, these values are the result of a retrospective application in 1938, of a salt correction factor of 1–35. The justification for this factor is examined. It is concluded that there is sufficient uncertainty in the accuracy of the pre-1948 measurements of phosphate concentration to preclude their use in long time-series analysis. A statistical analysis of the post-1948 data shows that a simple first order autoregression model provides a perfectly adequate description of the data; the data can be considered as essentially random, with a small component of carry over from one year to the next. The evidence is weak that variations in winter concentrations of phosphate support the Russell Cycle hypothesis.