Jones describes two problems to consider when choosing how to approach the test negatives. One is the changing lighting conditions over the period of time require to shoot a large number of negatives of different emulsions and over a range of exposures. And two, the spectral response of the different emulsions to the colors in the scene. In order to have the results of the judging relate only to the variables being tested, each negative has to be identical in both the illuminance of the scene and the spectral response.
I’m hesitant to bring this up because it will might distract from the point, but it’s necessary toward the understanding of the testing. Their approach not only solved the two problems but also provided a way to make very precise measurements. They shot the original scenes on black and white transparency film and make internegs from the transparencies. This way each negative was made from an unchanging scene without any issues of color.
Without going into detail, they took great pains to produce the internegs. From the transparency, they were able to make test series with variations in exposure and development. Here is an example of data from one of the test subjects, the famous Willow Pond. The subject had an average luminance range and consisted of many frequently encountered elements. The highlighted column shows the all the different average gradients. The column K is the gradient value at the print judgment speed point (first excellent print). The value of K is rather consistent and appears to be independent of the processing.
There were also tests with original scenes of shorter and longer luminance ranges. In A Study of Various Sensitometric Criteria of Negative Film Speeds, Jones discusses the results from a scene with a longer luminance range. “In making the series 9 to 12, a test object having considerably higher contrast was employed. This is designated as the “Woods Scene.” The brightness contrast of this case was 100. Here again the same four negative materials were employed. Again the values of K agree fairly well with those previously found, the average being 0.3196."
Last edited by Stephen Benskin; 03-13-2013 at 12:32 AM. Click to view previous post history.