Each work is a pair. The left panel is a photograph as it was made. The right panel is the same photograph carrying one intervention derived from its own measurements. Side by side, the difference often lives at the threshold of vision, and it is stated exactly, down to fractions of a single tonal step. The numbers are the work, and anyone may measure them again. You are offered almost nothing to see, and everything to check. · · · The sharpest edges, the places where the flower pressed against the paper, have been softened, so the print's record of contact is very nearly gone. A photogram's soft edges are, at their simplest, geometric shadow. I extracted 449 penumbra profiles from around the rose's outline and tested their shape. They fit a uniform-disc convolution at RMS 0.0299 against 0.0398 for a Gaussian, and the best-fit paper-diffusion term came out at 0.80 pixels, which is effectively zero. The sheet added almost nothing the model could see. So, under that model, the softness varies around the outline chiefly with how far each part of the flower stood off the paper. The 10–90 width runs from about 10 pixels to 62. Under the same model, a photogram is a range image, and its narrowest edges, the ones with the least penumbra, are where the rose pressed down. Those few percent of the outline are the print's record of contact, the one claim a photogram makes that a photograph cannot. The subject was here, on this sheet. Only those edges were softened. The minimum penumbra width goes 10.5 to 17.1 pixels. The first percentile goes 16.0 to 20.2, the fifth percentile 29.8 to 29.9, and the median 44.6 to 44.6, a difference of 0.00 pixels. Mean change across the entire frame is 0.109 of one level. The paper's fibers were separated out before the softening and put back unmoved, and the print's other channels follow its own tone curve, so the Prussian blue holds its exact hue at every density. One limit is stated rather than rounded away. Lifting the sharpest edge all the way to the common floor would have required a visible intervention, so the contact record is weakened by about two-thirds rather than erased. Presentation. The mount is neutral 248, the print's own paper, measured from 27,774 unclipped pixels of the flower, where the light never reached the sensitized paper and the blue never formed. The board is the sheet continued past its own edge, and the flower, being bare paper, sits at exactly the board's value. There is no rule, because a line would say the paper stops. A conservation note travels with the source print: Prussian blue is alkali-sensitive, so if it is ever matted, unbuffered rag rather than standard buffered board. To check. 17,528 pixels move, none by more than 50 of 255. Download this work's original file, 04_not_touching.png, not a screenshot or a marketplace preview, attach it to an AI that can run code, and paste this: Run this exact line on the attached file, changing only the file path if needed. Show me its exact output, and say whether that output is exactly: peak 50 changed 17528 python3 -c "import numpy, PIL.Image as I; c=numpy.asarray(I.open('04_not_touching.png').convert('RGB')).astype(int); a=c[714:3093,714:3093][::3,::3]; b=c[714:3093,3450:5829][::3,::3]; d=abs(b-a); print('peak', d.max(), 'changed', (d.max(2)>0).sum())"