Final answer:
The error in Dennis's data display is not specified by the provided options. However, histograms can show percentages and bars do not necessarily have to be vertical, so options C and A could be dismissed as errors. Overlapping intervals and separating males and females into different histograms could potentially be errors depending on how the data was handled.
Step-by-step explanation:
The error Dennis made in his data display for the smoking habits project is not directly stated in the options provided. However, based on the given reference information, we can infer what a correct histogram should look like and identify possible errors. A histogram can indeed show percentages, as explained in Figure A8 (c), where the vertical axis measures percentages of the total population. Therefore, option C is not necessarily an error. Additionally, bars in a histogram do not have to be vertical, as horizontal histograms are also acceptable, eliminating option A. It's important that intervals in histograms do not overlap, because each bar should represent a distinct range, so option B could be a valid error if Dennis' histograms had overlapping intervals. Lastly, having males and females in separate histograms is not inherently wrong, as comparing them can be done effectively with separate displays if the scales are consistent, thus option D may not be an error either.