I measure a response variable y at several values of the explanatory variable x, which is "time of measurement." Plotting log y against time of measurement looks
approximately like a straight line with a positive slope. I can conclude that:
the rate of growth of y is positive but slowing down.
I should also take the log of the time variable.
I should have taken the log of the time instead of log y.
since only logs of numbers greater than one are positive, the line best fitted to
the plot of y against time would have to have slope greater than one.
an exponential curve would fit well to a plot of y against time.