The length-independent growth indices used in an earlier communication to compare the second-year growth of East Anglian herring have been computed for each of the ages three to six, it first being demonstrated that seasonal growth has ceased by the time the fish are sampled, so that the current year's growth can be estimated. Because some partially recruited year-classes are involved a direct test of the effect of gonad production on somatic growth is possible, with the result that no gonad effect apparently exists. A comparison of growth changes over a period of some twenty-five years for all possible age-pairs indicates a discontinuity between the first and subsequent ages of growth but a close and increasing similarity in year-to-year growth fluctuations for older fish. This feature of the data is consistent with the increasing mixing rate of consecutive year-classes as they grow older; they come to share an increasingly similar environment and therefore would be expected to undergo parallel growth changes. An obvious environmental change affecting growth occurred in 1945 following a period of relatively constant growth conditions, and since that time growth at all ages has varied considerably. In the 1960's the variability is large and during this period extremely high growth levels have been achieved. In general terms this is associated with the decline in stock abundance that has occurred in recent years, although it is recognized that environmental changes have occurred independently. A density-dependent effect can be discerned as early as 1956. The growth indices can be used to follow the growth history of individual year-classes and, by combining them, to derive mean length-at-age indices; the apparent differential reaction of 3 year-old fish and of fish older than three to the increase in the copepod Calanus in 1950 is then accounted for. They can also be used to refute the suggestion that the growth changes in the early 195O's were generated by a length-differential mortality amongst prerecruits. Average patterns of growth derived from growth indices for a period when environmental conditions were relatively uniform indicate that first-year growth is less than would be expected from the pattern established later in the life history. Over the pre-recruit phase and for those ages during which year-classes become adult the pattern is remarkably regular, reinforcing the evidence of a lack of gonad effect, and calling into question a common assumption that in fishes, and in herring specifically, puberty disrupts or interrupts the growth pattern.