Thanks! An interesting way of validating their data on something that seems very difficult to quantify. Although I suppose the problem is there is an addition variable in the skill of the printer. One negative can produce very different prints, depending on who printed it and how.
Yes, there's always going to be that subjective factor, i.e., the skill of the printer involved in most realistic printing scenarios. However, that's not how the study was done. The scientific method in empirical research of that era was to try to control all possible factors, by either eliminating them or keeping them constant. This allowed them to focus only on one variable of interest, namely, print quality. For example, projection printing was rejected because it would introduce additional sources of variability, such as optical flare and collimated light. The prints were not made by a human being, but rather by a device they built for the purpose. All of that was done to eliminate subjective or highly variable factors, such as the skill or artistic intent of the printer. Here's a picture of the contact printer they constructed:
The only subjective factor as far as I can tell was the operator's choice of paper contrast: "the operator made a careful estimate, based upon his considerable experience in this field, of the paper contrast most likely to give a pleasing print." By employing a lot of advanced data reduction, tone reproduction, and numerical analysis (at least for the era), they were able to offer nuanced quantification of the contributing factors to excellent print quality and, in their own words: "to arrive at relatively simple relationships which might be expected to apply in general." In other words, answering the question of "why is this print of first-choice quality?" was not about asking the judges about their subjective, aesthetic choices and preferences, but, rather by identifying and quantifying these generalizable relationships. Today, this sounds like a very basic aspect of the scientific method, but back in the 1940s, in behavioral research, this would have been considered a relatively modern approach.