[After this article was posted, Tom McAvoy was kind enough to respond, and a lively, detailed correspondence ensued. A résumé is included at the end of this original post, so please make sure you read the whole thing to get the full balance of opinion on the subject.]
In a recent paper in Applied Optics,1 Thomas McAvoy has concluded that the intensity of the fluorescence of the Shroud at different places corresponds sufficiently well with Bob Rucker’s calculated intensity of a postulated neutron flux at those places to constitute evidence for a causal relationship. This could be powerful support for Rucker’s hypothesis.
However, the process of determining the relative UV fluorescence across the Shroud is by no means as simple as it might seem, as McAvoy readily acknowledges. The whole area of the cloth was covered by a theoretical 24 photos, in three strips of eight, each of which was individually processed, by a team led by Vernon Miller in 1978, and reported in the Journal of Biological Photography. 2 Unfortunately the original negatives are lost, so analysis has been carried out on the photos posted on the internet at shroudphotos.com by Gilbert Lavoie and Tom d’Muhala, themselves derived from 4 x 5 inch transparencies made from the negatives by Miller. As Miller seems to have made this set of transparencies using colour filters, McAvoy has tried to restore them to their original colour, as best he can, by comparing them to the photos published in Miller’s 1981 paper.
The exact details of McAvoy’s manipulations are less important than what he finally derived and published, namely three strips of eight photos (although three are omitted), shown in Figure 5 (the middle strip), and Figure 8 (the top and bottom strips).
And here they are again, extracted from the paper, with the middle strip between the other two.
With the best will in the world, we can easily see that this collection is far from uniform, both in colour and intensity. The two rows of red circles, and the two rows of green circles, mark exactly the same places on the cloth, but can be seen to be very different in the photos, and there are differences even horizontally across the strips, especially in the two right hand photos of the top and bottom strips, which are successively bluer and then redder than the others of the same strip.
By pushing the contrast and saturation, the differences can be seen more clearly.
Given this wide disparity in colour and intensity, I do not think it possible to conclude that all the photos were similarly processed. In a reply to my comment to that effect in a recent podcast, 3 McAvoy said: “[Hugh] is wrong about the image colour having any significant effect on my results. I have attached a copy of a paper of mine that was published in Applied Optics before the paper I discussed in the podcast. It demonstrates that image intensity contains essentially all the information in the uv images and that colour contains almost no information (see section 3). This paper refutes Hugh’s colour arguments.” I don’t think it does; in fact I think it misses my point. The differences in colour may have no effect on McAvoy’s consideration of the intensity levels, but they do demonstrate that each photo was independently, and slightly differently, processed, such that we cannot be sure that development times and temperatures, and the freshness of the chemicals involved, were the same for all the photos. In fact, we can be sure that they weren’t, since the same place on the Shroud appears quite differently in two different photos.
Thus, for example, We cannot be confident that the middle of the back fluoresced more brightly than the chest, even though the back photo is brighter than the front one, nor that the top strip fluoresced more brightly than the bottom one. Bear in mind that these two illustrations (enlarged from the image above) cover exactly the same area of cloth…
… and have been carefully manipulated to try to make them identical.
McAvoy went on: “Vern Miller, the photographer who took the uv photos of the Shroud, was a professional and it is doubtful that he would have processed each of his photos differently. In the paper that he published with Pellicori in 1981 that shows the uv images, he discusses the difference between the uv Shroud images and visible light images of the Shroud. Had either the uv or Shroud images been taken with variable lighting, or processed differently, as Hugh contends, such a comparison could not have been made. Lighting and processing were almost certainly not variable.”
I have to disagree, although I dare say that every effort was made to keep things the same. Surely McAvoy tacitly acknowledges this in his omitting three photos completely from his investigation, as they are obvious outliers. All 24 areas were photographed. In their paper, Miller and Pellicori discuss seven photos covering the central strip, and a detailed examination of some of the photos published at shroudphotos.com, by Sam Pellicori, includes one of the dorsal thighs, omitted by McAvoy. 4
Nevertheless, assuming his manipulations valid, McAvoy derives three intensity graphs from the these photos, as follows:
These lines are supposed to correlate to Bob Rucker’s predicted estimations of the radiocarbon age at various places over the Shroud, whose details can be found in a paper on academia.edu. 5 Rucker’s hypothesis predicts that close to the shelf-floor and back wall of the sepulchre, radiation effects were increased due the reflection of neutrons off the hard surfaces, so therefore the dorsal image (lying on the floor) would fluoresce more than the ventral image, and that the whole side of the cloth opposite the side-strip (the side against the wall) would fluoresce more than the side-strip side. The fact that he can see two maxima on two of his graphs, and that the top strip is brighter than the bottom strip is sufficient to convince McAvoy of a correlation, but there are a number of serious flaws to this conclusion, which close inspection will make clear. Here are McAvoy’s graphs again, and images of the photos to help the examination.
McAvoy’s correlations relate to five observations that he lists in his introduction.
1). “Average UV fluorescence is highest in the centre of the dorsal image on the Shroud.” While I agree that after McAvoy’s adjustments, the one photo showing the centre of the dorsal image is brighter than all the others, it is not clear that it truly represents the Shroud’s relative fluorescence. That photo significantly overlaps the one below, so the overlapping areas of both should show the same fluorescence, but this is not the case.
Here are the two relevant photos extracted from above (left), and the colour removed (centre). A small rectangle has then been copied from the top photo to the lower photo, and the lower photo brightened so that the two areas match (right). It is now much less clear that “Average UV fluorescence is highest in the centre of the dorsal image on the Shroud.”
2). “Except for one comparison out of 10, the dorsal side of the Shroud fluoresces more (has higher average intensity) than the frontal side.” Without adjusting the lower band, the dorsal side of the Shroud is represented by seven photos, whose intensities are 46.9, 47.3, 47.9, 48.4, 49.5, 50.7, 55.9, average 49.5. The ventral side is represented by eleven photos, whose intensities are 46.2, 46.4, 46.5, 46.7, 47.1, 47.2, 47.8, 48.8, 49.7, 50.7, 52.6, average 48.2. In this case, the statement is correct. However, if the whole of the lower band is corrected in the same way as the single photo above, the resultant image is on average brighter on the ventral than on the dorsal.
3). “Six out of seven of the top sections of the Shroud fluoresce more than the corresponding bottom sections.” Only if the bottom section is uncorrected, so that where it overlaps the top section, the same place on the Shroud appears darker on the lower strip than it does on the upper. If this anomaly is corrected, so that the same areas have the same intensity, the result looks like this:
Now, as might be expected, we can see that average intensity is much the same all over the Shroud.
4). “Along the centre images, average UV fluorescence intensity goes through two maxima.” This is true, but not saying much. Any random list of six numbers is quite likely to have two maxima, but, assuming that this is not random for a moment, the two maxima are firstly on the middle of the back, as predicted by Rucker’s hypothesis, and secondly between the ventral hips, which is not. We note that the upper strip also has two maxima, between the two heads and between the ventral hips, neither of which concur with Rucker’s hypothesis. The lower strip’s maximum is at the dorsal feet, which again, contradicts Rucker. Of the five maxima, only one is predicted by the neutron radiation hypothesis.
5). “Along the centre images, average UV fluorescent intensity, drops off sharply toward the feet.” The emphasis on the centre images suggests bias, in my opinion. Along both the upper and lower images, average fluorescence intensity actually rises towards the dorsal feet, and at the ventral feet, it may be that the thick strip of non-fluorescent backing material has affected the average UV intensity calculations.
So none of the five observations above can be said to match Rucker’s hypothesis sufficiently well for any conclusions about correlation to be made; and there is another aspect of the photos which also tends to contradict it. For Rucker’s hypothesis to be verified, there should be a variation in fluorescence not only between photos, but also within each one. This is not easy to evaluate, as each one is brighter in the middle and vignetted in the corners, but it can be checked. Take the sequence of three photos of the central series, from the middle of the back to the middle of the chest, which ought to show a drop in intensity from one to the other.
None of the rectangles demarcated on the lower set of photos corresponds to the intensity curve, sketched on the top set, that ought to be present, if McAvoy’s hypothesis of variable fluorescence is viable. The central photo of the ventral shins also shows no sign of the fluorescence dropping off sharply, or, in fact, at all.
For what it’s worth, here are all 24 photos, as I received them, with their filters applied, which I shall comment on beneath.
Bearing in mind that the centre strip overlaps about half of both the top and bottom strips, it is clear that the processing of each strip as a whole, as well as individually along each strip, is individual. It seems to me that photo 1C, 2B and maybe 2G are darker than they should be, and that 2C and perhaps 3D are lighter. We can see this more clearly by making an intensity/height model and looking at the strips sideways on.
If this demonstrates anything, it is that the Shroud fluoresces more or less evenly, and that no correlation between its fluorescence and any predicted neutron irradiation intensity can be shown.
===========================================================================
Thomas McAvoy responds:
Reply to Hugh Farey’s Comments on UV Photos.
Hugh Farey has carried out an interesting and detailed study on the Shroud UV photos that I discuss in 3 journal and 1 conference paper. Commenting on the Shroud UV photos Hugh states:
“With the best will in the world, we can easily see that this collection is far from uniform, both in colour and intensity. The two rows of red circles, and the two rows of green circles, mark exactly the same places on the cloth, but can be seen to be very different in the photos, and there are differences even horizontally across the strips, especially in the two right hand photos of the top and bottom strips, which are successively bluer and then redder than the others of the same strip.”
Let me start by stating that I completely agree with Hugh’s points on colour and intensity in this statement. I do, however, disagree with Hugh’s explanation for the differences in intensity and colour and that they might be caused by different processing of the film used to take the photos. Let me start by addressing intensity and then move on to colour and processing.
[McAvoy divided his response into two sections, labelled Intensity and Colour. The second will be discussed later.]
Intensity
1. How photos were taken: All the Shroud photos, UV, visible light, etc. were taken using the rail system shown below.
This rail system was used so that a large number of images could be taken. In this image, A designates the 200 watt xenon strobes, B an AC powerpack, C the camera, and D an adjustment rod to keep the distance to the Shroud fixed. The rail was 14 feet long and it allowed the camera to be moved down the length of the Shroud with pictures taken along the length. Having the rail system facilitated taking a large number of photos in a reasonable amount of time. It is my understanding from talking with Barrie Schwortz that for the visible light photos Miller took, an additional lighting source was used and that Barrie used a light meter to check that the Shroud was uniformly lit. For the UV photos the 2 xenon were turned to a 45o angle as shown below.
In taking the UV photos, the output from the xenon UV sources was filtered to eliminate visible light from the sources and the camera had a UV filter to eliminate reflected UV light. The resulting UV photos would then show the visible light effects produced by exposing the Shroud to the UV sources. When the camera was moved to a new position, the strobe lights would flash and an image would be taken and the camera would be moved down along the Shroud. The UV images that I excluded as outliers were much darker than the others. I asked Barrie Schwortz about this and he said that for these images he felt that the strobe lights probably misfired resulting in the dark images. Given the large number of photos taken, the misfiring was probably not detected when the photo was taken, but after the film was developed.
And Hugh Farey replies:
What do we mean by “misfired”? If the flashes did not go off, then no image would be seen at all. If only one flash went off, some directional information would be visible on the image. If the Italian voltage was variable, then different intensities could be seen all over the Shroud. Perhaps they are.
2. Effect of non-uniform lighting: For the UV photography supplemental lighting was not used. As a result, there was what I called a hotspot of incident UV light on the Shroud where the beams from the 2 lights converged. I discussed this point in my 2019 Applied Optics paper. The image below is taken from that paper.
This image shows a contour plot of the image intensity for one of Miller’s UV images (image 3d-UV-S1-E-12_0476 on shroudphotos.com). The hot spot is clearly shown in this contour plot. All of the contour plots for Miller’s UV images look like this one. This contour plot demonstrates that unlike the visible light images, the lighting for the UV images was not uniform. This lack of uniform lighting has to be considered in doing any analysis of Miller’s UV photos, and Hugh did not take this into account.
Yes, I did. The problem of ‘fall-off’ is an important one, and to compare the intensity of the same spot in different photographs, it is important to choose a spot which has received the same intensity of light on both images.
If I take a UV image of the same object using 2 different UV source intensities, I get 2 different intensities for the resulting UV image which is in the visible range. For example, if 1 UV source had twice the intensity of the other, each pixel at identical spots on the object photographed would have an intensity ratio of 1:2 when comparing one photo to the other. Since the red and green circles in Hugh’s diagram were exposed to different UV source intensities they cannot be compared in terms of their intensities. These circles can only be compared if the UV intensity that they received is known, and it is not. By contrast, since all the UV images except those at the end of the Shroud are the same size and the section of the Shroud photographed was exposed to the same but non-uniform UV light source, it is possible to compare the average intensities of the images to one another. This is exactly what I did, and the patterns I found agree with Bob Rucker’s neutron simulation results. Much of Hugh’s analysis is based on comparing overlapping UV areas to one another but because of the non-uniform UV lighting intensity such comparisons are not valid.
HF comments: This is an important point, and I most certainly considered it carefully. Consider the pair of photos showing the back of the man on the Shroud. Here are McAvoy’s contour plots (annotated by me).
The first image shows the two plots superimposed, and the overlapping area shown with a black rectangle. The second pair show the individual images. Areas A and B are exactly the same place on the Shroud, but I certainly accept that because A is close the to middle of the photo, and B close to the outside, it is not fair to expect them both to be at similar intensities. However, sites C and D are also exactly the same place on the Shroud, and this time they are also in a very similar place in terms of the the “hot spot” in the centre of each photo. If the circumstances of the production of both photos were really exactly the same, C and D should certainly be the same colour. That they are not is powerful evidence that the circumstances of the production of the photos were not identical.
The colours in this image represent image intensity, and they have nothing to do at all with the colours in the UV images themselves. In the CIE Lab colour space colour is independent from intensity. The colours in the above images show intensities and the intensities are from highest to lowest white-gold-green-blue. As the plot shows there is an enormous difference between the intensities of points A and B, order of a 3 to 4 fold difference. For point C and D the difference is less but probably on the order of 1.5 to 2 fold. The image intensities at these 4 points can easily be calculated in the CIE Lab space to determine the exact differences. Since the points being compared are the same, these intensity differences are caused by the difference in the UV light that each point received. It is not correct to say that the C and D points “are also in a very similar place in terms of the “hot spot” in the centre of each photo.” So, this image reinforces my point about why intensities at the same point on different images cannot be compared. Consider taking an ordinary photo of a scene at mid-day and at dusk with exactly the same camera settings and position. The intensity of the image in these 2 scenes would clearly be different. Comparing the intensities at points A and B and C and D is the same as trying to compare the intensities in the 2 scenes at different times of the day.
Why does McAvoy disagree that points C and D are in similar places in terms of the “hot spot”? If the lighting conditions were the same for both images, then the intensity of the UV light falling on each image varied exactly similarly from a maximum in the middle to a minimum in the corner, and the fluorescence of a single spot illuminated by light of the same intensity should be the same. But it isn’t. Here are our photos, whose colours, loosely delineated with ‘isochromes,’ indicate the fluorescence intensity. A and B are the same place on the Shroud, but, being illuminated with different intensity, fluoresce with different intensity, and are shown in in different colours, but C and D are not only on the same place on the Shroud, but also within the same isochrome. If the original photos had been taken under the same conditions, C and D should fluoresce with the same intensity, but they don’t.
McAvoy in fact, perhaps unwittingly, seems to agree with me when I say that the film, exposure, or processing must have been different for the two pictures. In his case, he explains that two scenes look different under different lighting conditions (midday and dusk). Exactly. The two ‘scenes’ on the Shroud were also taken at ‘midday’ and ‘dusk’; the right hand one must be lightened – and so must all its fellows on the bottom row of the UV photos, in order to provide a fair comparison.
The 2 images that Hugh analyzes appear to be taken from Figures 11 and 14 of my 2019 Applied Optics paper. In his reply Hugh questions whether the lighting conditions were the same for the two images. He makes an interesting point, and indeed there is a difference in the fluorescence intensity at points C and D. There are two reasons for this difference, one somewhat minor and the other significant. The 2019 paper states that the image on the left was not centered, while the image on the right was. Thus, one cannot simply take the elipses that Hugh has drawn for the figure on the left and superimpose them on the figure on the right. The hotspot occurs in the center of the figure on the right since it was centered and toward the right of the figure on the left. However, moving the ellipses on the figure on the right so that they start at the hotspot would only contribute slightly to the difference in fluorescence intensities.
I agree with some of this. Looking at the two photos with more extreme manipulation, we can more clearly see the centre of illumination of each one, and they are different.
But, as McAvoy says, this does not really explain the different illumination of the same spot in each photo.
A major reason for the difference in fluorescence intensities between points C and D involves the UV images that were analyzed. In 2021 I published a second paper that discussed the orange color of the UV web images. In that paper it was shown that the orange color was the result of using color correction magenta and yellow filters. The UV web images analyzed in the 2019 paper resulted from photographing transparencies of the original STuRP UV images through these color correction filters. Unfortunately, the original unfiltered STuRP uv images have been lost. The original D UV images down the center of the Shroud were published by Miller and Pellicori in 1981. The original E and B images were never published. By comparing the published D images with those on the web, I was able to develop a model of the color correction filters used in the production of the web images. The figure below taken from my 2021 paper shows the original web D8 image, its published version, and my corrected versions.
I can’t see any difference between (c) and (d) myself. Is there any?
Figure 1 (a) shows the web D8 image, (b) the 1981 published D8 image and (d) my reconstruction of the D8 image using my filter model. I used the model of the color correction filters to reconstruct all of the UV images used in the original 2019 paper.
Now let me address how the color correction filters alter fluorescence intensity. In the 2019 paper the average intensity for the left side of the image analyzed by Hugh was 44.17 and that for the right side was 27.56. The figure below taken from the 2021 paper shows fluorescence intensities for the color corrected Shroud UV images.
As can be seen the average intensity for Hugh’s left image (8 on x-axis) changes to roughly 56. For Hugh’s right image (9 on x-axis) the change is to roughly 47. The difference in intensities for Hugh’s left image is 56-44=12. The difference in Hugh’s right image is 47-27.5=19.5. Thus, the color correction filters decreased the intensity of the right image compared to the left image by an additional 60%. The uv intensity of the bottom B images is affected more by the color correction filters than the center images. I contend that the color correction filters are the major reason that the intensities at points C and D in images analyzed by Hugh differ. The STuRP photographers were skilled professionals and they would have paid attention to lighting details. In order to do the comparison that Hugh proposes one would need to have access to the original UV images. Unfortunately, these original images are not available.
It makes sense that adding a coloured filter to a photo means it not only changes colour, it gets darker. Removing the colour filter restores the original colour, and make it brighter again. McAvoy may be right about all this, but the whole point of his paper was to compare UV fluorescence from photo to photo. Unless overlapping photos can be adjusted so that the fluorescence from the same spot, illuminated in the same way, appears the same on both photos then the overall fluorescence of the two photos cannot be compared.
My contention is that if we had the original UV photos, then the same spots in 2 different photos would have the same fluorescence intensity, if the 2 spots received identical impinging light. My color correction filters resulted in images that were close to the original photos in terms of their color. However, there is clearly some error in my filter model. Secondly, without a measurement of impinging light intensity or model of this intensity it is not possible to know the intensity of the impinging light on a point on the Shroud. One can assume that of the 45-degree lighting is elliptical in nature, but its exact characteristics as well as what I call its hotspot center are unknown. I don’t see how you can determine the impinging lighting intensity from a single Shroud image. The intensity at each point is produced by 2 unknowns, the impinging light and the inherent intensity of the Shroud. One can certainly locate 2 identical points on 2 different images of the Shroud, but without knowing the incident light one cannot conclude anything about their fluorescence intensity. This is the reason I compared the average intensity of each image.
I think this was a significant difference between us. I do not care what the impinging intensity was. Its value is not important. My contention is that if the lighting set up was the same, then there are points in the overlapping sections of two photographs that received the same lighting, and therefore should have produced the same intensity of fluorescence. I illustrated this point with an illustration.
Left: two photos, corrected to uniformity by McAvoy, showing the overlap.
Bottom two (dark) photos: in exaggerated contrast, the overall shape of the impinging light can be inferred (blue ellipses).
Above them: as points C and D are both on the edge of the ellipse, they should be equally bright, but C is brighter than D.
Far right: Circle C has been moved to position D, and the photo lightened to match it. Now the two overlapping photos can be compared, and it turns out that overall their fluorescence is much the same.
You seem to use distance from the hotspot to infer the incident light intensity. Since the light source produced an elliptical-like incident light both the distance and the angle from the hotspot are required. Further, both light source intensity as well as inherent Shroud fluorescence contribute to fluorescence intensity at a point. The UV intensity around individual points can change drastically due to such things as scourge marks, blood and fluid on the Shroud. To compare a point you would need a very precise location as well as the input light intensity at the point, and this input intensity is unknown. I have attached a filled in contour plot of my color corrected D8 center image.
This image was generated with the Matlab contourf function. If you look at it you can see how complex the fluorescence pattern is. The pattern is elliptical in nature but the elipses are far from uniform in color. Your circles for the points you compare are much too large and they would encompass too much of a change in fluorescence intensity. Using them does not produce “comparable points” for comparison.
The fluorescence pattern is indeed complex, and a feature of the complexity of the Shroud’s markings – two different materials of patch, blood flows, scourge marks, etc. However, the overall pattern is very easy to make out, and this is a feature of the incident UV light. The image above is divided into levels of brightness, which can be treated as contours of equal brightness, and the pattern of input brightness reasonably inferred.
McAvoy was probably right that points C and D above should not only be the same distance from the centre of the hotspot, but also the same angle, so I reassessed my previous diagram.
The centre of illumination, I think, and its overall ellipticity, can be found reasonably accurately by adjusting the intensity and contrast of the image. This, I thought we agreed, was more or less the same for each photo, with the possible exception of the outliers you remove from consideration. This means that the same pattern can be superimposed on all the images, as illustrated in pale blue. The places where those patterns intersect are identical distances, and identical angles, from the hotspots, and received identical illumination. Three pairs of similar spots are indicated (red dots), and areas around them outlined (purple circles). The equivalent places on each photo should be the same brightness, but the lower one is darker. Before an objective assessment of the brightness of the cloth at different points can be compared, the intensity of one or both photos must be adjusted so that these particular strips are the same brightness.
Without that reconciliation, the average intensities of the photos of each strip cannot be directly compared, and the three lines on your graph, representing the average intensities along strips E, D and B, describe the intensities of the strips of photos, not of the equivalent strips on the Shroud.
I can however already hear an objection. Maybe the intensity of each strip falls off more in the “upwards” direction than it does in the “downwards” direction. To check if this is true or not, we can carry out exactly the same process as before, but with photos from strips E and D instead of D and B. This time the upper strip (E) is darker than the lower D), showing that the illumination does not fall off more “upwards” than “downwards.”
This may all look like a hammer to crack a nut, as to me it is perfectly obvious that the three strips are, each as a whole, of different brightnesses, but this demonstration makes it objectively clear, I hope.
Replying specifically to McAvoy’s points above:
“You seem to use distance from the hotspot to infer the incident light intensity. Since the light source produced an elliptical-like incident light both the distance and the angle from the hotspot are required.” Quite true. The new attachments cope with this, I think.
“Further, both light source intensity as well as inherent Shroud fluorescence contribute to fluorescence intensity at a point. The UV intensity around individual points can change drastically due to such things as scourge marks, blood and fluid on the Shroud.” Also quite true, or the entire cloth would simply look like a series of ellipses under UV light.
“To compare a point you would need a very precise location as well as the input light intensity at the point, and this input intensity is unknown.” Not so. You do need the point to be quite small so that ‘drop-off’ is reduced to a minimum – and I have chosen much smaller ones than previously in my illustrations above – but as long as they are in the same place, they contain the same imperfections, blood, image, scorch etc. You do not need to know what the input light intensity was, only that it was the same for that spot on both photographs.
“I have attached a filled in contour plot of my color corrected D8 center image. This image was generated with the Matlab contourf function. If you look at it you can see how complex the fluorescence pattern is. The pattern is elliptical in nature but the ellipses are far from uniform in color.” True but not relevant. The ellipses are generated to describe the input illumination, not the output, which is of course highly variable. The input sources were simple, and the incident UV light can be modelled as simple ellipses. if the lighting set-up wasn’t altered, the eccentricity of the ellipse should be the same for each photo, and its position on each photo achieved by adjusting the intensity and contrast. There is a certain subjectivity to this, but not, I hope, suffiencient to invalidate my conclusions.
“Your circles for the points you compare are much too large and they would encompass too much of a change in fluorescence intensity. Using them does not produce “comparable points” for comparison.” I agree with this, and have reduced the size of the circles in the illustrations above. I think they are not now too large.
“You have not answered my question about why the averages over each image cannot be compared.” I think I have shown that the central (D) strip is brighter than the other two, not because the cloth was brighter, but because the photograph itself is brighter. This is shown by the difference in brightness of the same places in the different strips, where they should be the same. By looking at my illustrations above, strip D is brightest, strip E a little less bright, and strip B less bright still. This is what is reflected in the different average levels of the three lines on your graph, not any real difference in fluorescence.
I do not agree with this statement: ““The centre of illumination, I think, and its overall ellipticity, can be found reasonably accurately by adjusting the intensity and contrast of the image.” You seem to be saying that somehow you can determine the input light intensity of an image, and then superimpose it on all images. Specifically, you state: “The ellipses are generated to describe the input illumination, not the output, which is of course highly variable.” I disagree that this can be done. Since, the intensity of a point on an image is the result of both the input light intensity (variable 1) and the inherent output Shroud intensity (variable 2) it is not possible to separately determine only the input light intensity (variable 1) from an analysis of a UV image. The intensity of a point is the result of a combination of the 2 variables. The UV image is the result of the input UV signal being absorbed by the cloth, and then producing an output color shown on the image. In determining your blue lines how do you eliminate the unknown Shroud inherent intensity?
Also, exactly what type of pattern the 2 45o lights produce is not known precisely. It is elliptical in nature, but one would have to model this 45oarrangement to determine the exact input pattern.
I thought this was missing my point, although I appreciate that I was also missing McAvoy’s!
I think you’re a bit hung up on my being able, or not, to “determine the input light intensity of an image, and then superimpose it on all images.” I don’t know the input light intensity, and I don’t want to determine the input light intensity, and I don’t think the input light intensity is necessary for my argument. The only thing that is important is that the input light intensity pattern was the same for all the photos. From the diagrams in Morris & Pellicori we can deduce that the incident light was in the form of an ellipse, long axis longitudinal to the length of the cloth, with a central area of maximum brightness, falling off to darkness at the edges, more rapidly in the vertical than in the horizontal direction. On a perfectly uniform surface, a similar reflectance pattern would have been returned wherever the set up was pointed.
The fact that all the photos do in fact show an elliptical reflectance pattern as described is evidence that the input pattern and output patterns are related. From all my diagrams, it is clear that some areas of the Shroud were photographed twice, and that some parts of those areas must have received the same input pattern, and therefore should show the same output fluorescence. The exact shape of the input pattern isn’t very important: somewhere they overlap. That they don’t show the same fluorescence is evidence that in fact the original assumptions are not valid. Either the lights and camera were not identically placed for each photo, or the electrical input varied, or the exposure time varied, or the photographic film was not wholly uniform, or the developing chemicals varied, or the developing time varied, or the subsequent printing process varied, or any number of combinations of those or other possible factors. This means that although the fluorescence of different areas on the same photo can be compared, it is not possible to say that the fluorescence of one photo can be compared to that of another, or that on the Shroud itself the overall fluorescence of the area delineated as strip E is truly greater than that of strip B, even though it looks greater on the photographs.
I think that what you are doing is mathematically impossible. The intensity, I, of each pixel on a UV image is a function of the intensity of the UV light that impinges on it, x1, and the inherent ability of the Shroud to emit fluorescence at that pixel, x2. Mathematically I = f(x1,x2). This function could be inverted to give x1 = g(I,x2). From the image the intensity, I, can be accurately determined at each pixel. To determine x1, or any property of x1 it is necessary to know x2 which you do not know. A property of x1 would also be a function of both I and x2. Your elipses seem to be an attempt to determine either x1, or a property of x1 without knowing x2 and that is mathematically impossible.
Also, I am not sure just which images you are trying to analyze. If they are copied from the thumbnails given in my 2 papers, then these are very low-resolution images compared to the very high-resolution web images that I analyzed. Each pixel in the images in my papers would effectively be an average of a number of neighboring pixels. The number would be determined by the size of the thumbnail images relative to the size of the high-resolution images. For the high-resolution images, intensity, I, can change significantly by moving only a few pixels.
===========================================================================
The discussion continued. Tom McAvoy based his counter-arguments on the necessity to consider the smallest possible areas for comparison of intensities, and on the fact that converting an image to greyscale does not accurately reflect the intensity of the colour original. However, taking one of his high resolution images, desaturating it, and comparing the intensities in ImageJ shows that the intensity is not changed by changing it to black and white.
We both found this confusing, and although I found it satisfactory, McAvoy did not. He informed me, quite correctly, that when an image generating program converts from colour to grey scale, it does so with the following intensity correction factors: 0.299R+0.587G+0.114B. We compared ImageJ, MacBook .pages, Adobe Photoshop and Matlab, and they all produced similar results. This suggests that pale areas, made of a mixture of red, white and blue, should become darker than bright red areas of the same apparent intensity, such as the bloodstains, while yellowish areas, made of more red and green than blue, should become less dark, as the green is the least reduced by the formula above. Sure enough, if we convert a saturated disc into greyscale that’s what we get:
This puzzled me for a while until I realised that the level 255, the maximum for each colour, is not a measure of intensity after all. The Red, Green and Blue sectors in the diagram above are made of each of the three colours given its maximum value, but they are not all of the same intensity, as we can see using ImageJ (or any other image manipulation program):
For this reason, the factors involved in the greyscale conversion accurately preserve the actual brightness of the original image, and all my comparisons above are perfectly valid.
The bottom line is that it cannot be demonstrated that the UV images have been produced under consistent conditions, and that therefore, no comparison of intensity between them can be said to reflect actual differences of intensity on the Shroud.
===========================================================================
[This is the Colour section of McEvoy’s original response to my post. We did not discuss it specifically in our subsequent exchanges, but my responses are in blue.]
Colour
1. Analysis: In section 3 of my 2021 Applied Optics paper, I discuss the use of Principal Component Analysis (PCA) to analyse the information content contained in an image. PCA is a statistical data reduction technique that looks to find combinations of variables that explain decreasing amounts of the variance in a data set. PCA can be used to assess what variables contribute to the information content of an image. In using PCA I first converted Miller’s UV images to the CIE Lab colour space shown below.
In this colour space there are 2 colour dimensions (a,b), shown in the disk, and an intensity dimension (L), shown vertically. L, a, and b are orthogonal and they contain different information about the image being analyzed. The mathematics of using PCA to analyse Miller’s UV images is given in my paper, and I don’t want to repeat it here. What I show is that the intensity variable, L, captures greater than 96% of the information in Miller’s UV photos. The remaining information which is less than 4% is contained in the 2 colour variables, an and b. There are a number of reasons why colour differences could occur in Miller’s UV images, but their information content is very small. For example, Miller’s UV source filter allows light between 400 and 410 nm to pass to the camera. Light in this range is in the blue visible spectrum. In essence PCA shows that colour is essentially unimportant in terms of the information contained in Miller’s UV photos. What is much more important is their image intensity. Differences in the colour of Miller’s UV images are not nearly as significant as differences in their intensity.
I can’t agree with this at all. The lower series of photos is a different colour from the middle and upper series. There must be a reason for that. Maybe the film was slightly different, or the lighting voltage during exposure was different, or there was something different about the processing, or some other factor. That difference ‘on the ground’ seems to have made a difference to the colour, and it may have also have made a difference to the apparent intensity. As McAvoy says, the two aspects, colour and intensity, are orthogonal and therefore unrelated, but that does not mean that the cause of the colour change could not also have been the cause of the intensity change. I think that McAvoy’s own contour plots show that the intensity difference is significant.
Hugh places a great deal of emphasis on the colour of the UV images. What my 2021 paper shows is that colour contributes almost no information in the UV images. One can use principal component reconstruction to demonstrate this point. Consider the figure below.
The image on the left is the original D8 image shown in my 2021 paper. This image has 3 independent variables and therefore principal component analysis of it yields 3 principal components. The image in the middle of the figure shows the results of using only the first principal component (PC1) to reconstruct the original image. As discussed in my 2021 paper, the first principal component is essentially the image intensity. The image on the right shows the results of using the second and third principal components (PC2+PC3) to reconstruct the original image. If all 3 principal components are used together (PC1+PC2+PC3) to reconstruct the D8 image, one gets the original image on the left. The last 2 PC’s essentially contain only colour information. The above figure clearly demonstrates that essentially all the information content in the original D8 image is contained in its intensity, and almost none is contained in its colour dimensions. This same conclusion is also true for all of the UV images. In effect one can consider colour differences in the UV images as being similar noise. Hugh is taking the variable that has by far the smallest effect on image information content and magnifying its importance to being the most important variable. As I stated in my earlier reply, colour contributes less than 4% of the information content in UV Shroud images. Intensity contributes greater than 96% of the information. Focusing on colour overlooks the important information that is contained in the intensity of the Shroud images. These PC results demonstrate that colour in these images does not provide any “powerful evidence” at all about the images, or their production.
I disagree completely, although I appreciate McAvoy’s calculations. The fact is that the bottom strip of photos is a different colour from the top strip. Rather than pretend it isn’t, I think it fair to enquire the reason for this. Is it random? I think not. I think the bottom strip was, in some way and at some stage, produced differently from the top strip. And, as the comparison with the middle strip demonstrates, the difference in production, however minor, has also made that strip appear to show less fluorescence from the same part of the Shroud than the middle strip.
Processing of images
1. Discussion: Vern Miller, the photographer who took the UV photos of the Shroud, was a professional and it is doubtful that he would have processed each of his photos differently. Reference 6 discusses how the STuRP photographers went about taking their photos and the detailed preparations that they made. I have had numerous discussions with Gil Lavoie about Miller, whom he knew, and Gill attested to Miller’s professionalism. The paper that Miller published with Pellicori in 1981 shows the UV images, and they discuss the difference between the UV Shroud images and visible light images of the Shroud. Had each of the UV or Shroud images been processed differently, as Hugh contends, such a comparison could not have been made. Processing was almost certainly not variable.
I cannot agree with this. If the films, exposure and processing were all the same, then the colour of overlapping areas would be the same, but it isn’t.
As Hugh points out Miller’s original UV negatives were lost. What was posted on the web were 4×5 transparencies made from the negatives and these transparencies had a distinctly orange hue. This hue likely resulted from Miller experimenting with different colour filters on his photos, in this case a magenta and yellow filter. However, the orange hue does not affect the PCA analysis which is identical for both the orange hue photos and those in my 2021 paper.
What I will be happy to concede is that the original negatives may have been much more consistent than the extant reproductions appear to show. I have attributed the differences we see to different conditions before exposure (slightly different film, films of different ages or kept at different temperatures, or even X-rayed as they went through customs), or during exposure (most probably due to variations in the electricity supply to the lights) or after exposure, which I have called “processing.” I would be happy to split that into the actual processing of the film in Turin, and into the “post-processing” processing, resulting in the photos as we have them today. It may be that only the last is actually responsible for the differences.
===========================================================================
This revised post is the result of a very detailed, very closely argued series of interchanges between Tom McAvoy and myself. Throughout, McAvoy was patient and kind in his attempts to explain to me exactly why he had come to his conclusion, and although in the end, as you can see above, I was not at all convinced, I learnt a lot about colour manipulation and valued his responses. If only all conversations with authenticists were as harmonious!
===========================================================================
1). ‘Shroud of Turin Ultraviolet Light Images: Colour, and Information Content,’ Thomas McAvoy, Applied Optics, August 2021.
2). ‘Ultraviolet Fluorescence Phototography of the Shroud of Turin,’ Vernon Miller and Sam Pellicori, Journal of Biological Photography, July 1981.
3). ‘Shroud Wars: Re-Evaluating the 1988 Carbon-14 Dating,’ Real Seekers, 12 April 2024.
4). ‘Image Analysis of the Miller & Pellicori UV Fluorescence Images of the Turin Shroud,’ Samuel Pellicori, shroud.com, December 2020.
5). ‘The Carbon Dating Problem for the Shroud of Turin, Part 3: The Neutron Absorption Hypothesis,’ Robert A. Rucker, academia.edu, July 2018.
6). ‘Quantitative photography of the Shroud of Turin’, Donald Devan and Vernon Miller, IEEE 1982 Proceedings of the International Conference on Cybernetics and Society, October 1982
7). ‘Analysis of UV photographs of the Shroud of Turin’, Thomas McAvoy, Applied Optics, September 2019.
8). CIE Lab Color Space, https://en.wikipedia.org/wiki/CIELAB_color_space.