If case weight falls within a fairly narrow and consistent range, adding the extra weight of water plus the case shouldn't change the slope of the line, only the y-intercept. Case volume will always show the general trend of decreasing as case weight increases. Your first method of plotting the data is misleading with regard to the effect of case weight on case capacity. It's merely a way of making the data look much better than it really is because the critical value, which is the actual water weight or case volume, becomes proportionally reduced when the weight of the case is also included. For example, take the highest (~90 gr) and lowest (~84.7 gr) case weight values. When graphed appropriately to illustrate the inverse relationship between case weight and case volume as shown directly above, these values clearly show up as outliers. In your original graphing approach, they would be spot on a trend line, if one were present, which is misleading because the trend line that is implied by the original scatter plot you showed would have a positive slope. Perhaps it's only my impression and maybe you didn't mean to imply that, but the original graph certainly gave me that impression. Case volume will always show a general decrease (negative slope) as case weight increases, regardless of whether the weight of the case is included or not.
You are correct. On every count. The original graph can be misleading.