I would be inclined to say that your hypothesis was partially correct. I wonder, however, if perhaps your hypothesis isn't a conglomeration of two separate hypotheses. The first, predicting a linear trend, is supported by your data, while the second, predicting a plateau, is falsified.