In his 1951 study of Nile River data, H.E. Hurst introduced the rescaled range statistic-the image statistic. He argued via a small simulation study that if image, image, are i.i.d. normal then the image statistic should grow in the order of image. However, Hurst found that for the Nile River data, the image statistic increased not in the order of image, but in the order image, where image ranged between image and image. He discovered that the effect also appeared in other sets of data. This is now called the Hurst phenomenon. We shall establish some unexpected universal asymptotic properties of the image statistic, which show conclusively that the Hurst phenomenon can never appear for i.i.d. data.