Appendix II
Analysis of a Paper Photograph

Occasionally one encounters an old photograph that is different in some respect from all the standard types in our memory, either personal or computer. The frequency of such encounters is a function of the experience process: there is always something new to be learned. Following is the story of the casual investigation of a photograph that puzzled this writer for several years in spite of diligent literature searches. I hope it sheds some new light on a topic that was found to be very sparsely documented.

Figure 15 shows an unframed tinted portrait whose actual dimensions are 16 x 20 inches. It is on rough matte paper glued to coarse cardboard; the paper is 0.0087 inch (0.22 mm) thick on 0.035 inch (0.88 mm) cardboard. It is tinted in at least three colors, and the paper and cardboard are yellowed and crumbling. The photographic image was barely perceptible and evidently served only as a guide to tinting. There are no identifying marks on front or back, but it was known to have been made in Columbus, Ohio in 1901 plus or minus one year; the date and location are known because the subject is the mother of the author. Figure 16 shows a small mounted print obviously from the same negative that was printed on conventional contemporary paper, untinted.

Figure 15 Figure 16
The question is what type of exposed-fiber photographic paper was used, and what was the sensitizing process. Bromide enlarging paper was widely used by the date of the photograph, and is easily identified by the baryta-undercoated emulsion. Some practitioners were still using albumen paper, but this also is easily identified. Presumably the photographer used fiber paper because, being rough, it was easier to tint, either with water colors or Conte crayons or other media.

The FOTOFIND program (Chapter 14, Section 4) was used to list all the paper processes with exposed fibers (no emulsions). The result is shown in Figure 17, including responses to the questions. Note that 'n' was answered to the question about retouching; if we had answered 'y'or 'u', the program would have returned 'crayon print' as the search result. We answered 'n' because we were trying to list possible uncoated processes. Since the subject photograph is a commercial product from the photographer who produced the table portrait, only the first six candidates need be considered. Cyanotype can be discarded: it was the result of answering 'uncertain' to the color question. If we discard platinotype and palladiotype because of the high cost of a 16" x 20" picture (unneeded cost because of subsequent tinting), we are left with calotype, kallitype, and diazotype.

Figure 17

There was an additional consideration: a gum-bichromate print, under-exposed and almost totally washed-out, which would have exposed most of the paper fibers (gum prints were popular in the 1890's). But this is improbable because it is clearly an enlargement and bichromate processes were too low in sensitivity for enlargers of the day. Microscopic examination of this print at 90x failed to show an emulsion.

Conversations with archivists of four museums revealed that they, too, possessed similar portraits, some of them charcoaled rather than tinted. In at least two cases the subjects were of historical interest. In conversations with this writer, none of the museum personnel could identify the process or the dates.

I found two other similar family portraits 14 x 17 inches in size that had a monochrome brownish color. Matching copies on cabinet cards were also found that were obviously made from the same negatives. The cabinet cards appear to be made on conventional silver chloride paper and showed some tarnishing, but the large prints did not show tarnishing.

The most obvious explanation was that the photographic process consisted merely of an under-exposed silver print to give the illusion of free-hand art work. An experienced dealer in 19th century photographs was consulted, who made the plausible suggestion that the pictures may have been printed on a thin diluted emulsion hand-coated by the photographer. But the failure to find traces of emulsion at 90x was puzzling. It was a reminder that there were many private process variations in the 19th century, not all of which were publicly documented. However, library searching failed to turn up any mention of such work, which was inconclusive.

Closer examination of the center of the large print at higher magnification was needed, to search for traces of residual emulsion. For this, and other work, we wanted to examine all regions of these pictures at 200 - 300x. We modified the mount of a biological microscope to permit inspection of the centers of such large prints. With this new capability it was possible to see faint shiny traces in scattered locations in the center of the image, but no coherent or continuous layer. The examination did not establish whether the tints were water colors or pigments. There were faint traces of highly diluted color that had no discernible grain, but there also were some larger clumps of color.

The crumbling of the paper provided several loose or semidetached flakes. My colleague James Thrall examined and analyzed two of these flakes by x-ray fluorescence in the scanning electron microscope described in Chapter 14 and Appendix I. By this analysis it was hoped to determine the nature of the sensitive material. A quantitative analysis of two loose flakes showed the following results:

Table 4

Sample #1 Sample #2
Element Weight % Weight %
Si 16.39 13.28
S 12.35 11.36
Cl 1.48 ND
Ca ND 4.86
Fe 8.48 6.77
Br 20.65 14.18
Ag 12.67 9.62
Sb 14.91 ND
Ba 13.06 14.23
Pb ND 25.70

_____ _____
Total 99.99 100
(ND = Not Detected)

Conclusion: the sensitizer was probably silver bromide. Chlorine was low, eliminating the salt print or calotype process. The chemical elements in diazotypes could not be detected in the SEM, and the silver content that was found eliminates diazotypes. The iron content could be indicative of the kallitype process; the SEM analysis suffers from an artifact iron peak, which probably did not entirely account for the reported iron percentage. Pernicano [115] gives several fomulae for coating modern kallitype paper that include silver and iron; one method also uses barium. But as we shall see later in this Appendix, some of the components in the analysis are probably from tinting pigments, including iron.

If a known kallitype print had been available for calibration, it would have been helpful. But there were several variations of the process, and a single analysis will not be conclusive. It was fairly common for workers to sensitize their own paper with the kallitype process during this period; there are more details in Chapter 2 and in the references.

Since the electron microprobe generates x-rays from a very small sample volume (a few cubic micrometers), the quantitative percentages are probably not representative of the image macrostructure or the sampling sites. The precision is likely to be no better than two significant figures at best, and can only be improved by more sampling.

Our analysis showed the atomic ratio of silver to bromine to be about 1:2 in both samples. Silver bromide, AgBr, has an atomic ratio of 1:1. Normally, exposed silver bromide is reduced to metallic silver during development, and the unexposed silver bromide is removed by hypo. This should leave a surplus of silver relative to bromine, instead of the 1:2 deficiency we found. If the silver in the image had been selectively removed by a chemical treatment before or after tinting, it could account for the deficiency. To verify this, it would be necessary to analyze more sites in the portrait to be sure of representative sampling. Our evidence is suggestive, but verification by other workers would be desirable.

The most probable interpretation is the kallitype process, coated by the photographer. The chemical elements in the tinting compounds, and the lack of a known sample for comparison, leaves the identification tentative.

Scanning electron microscopes are widely used in industrial and academic research applications, but time on the instruments is not inexpensive. Our analysis was a volunteer effort performed by a good friend and colleague (see acknowledgments) who donated two noon hours. The information on the unexpected elements was a bonus. Analytical work frequently yields information that raises new questions, but one has to stop somewhere.

In Chapter 11 I have described what little I have found on "crayon prints" in the literature. Darrah [39, 43] describes tinting, especially the use of water colors and liquid aniline colors. These are organic dyes that would not have been detected in our microprobe analysis. Darrah [40, 192] is a more relevant reference. It describes crayon portraits that were reworked with ink or pencil, followed by removal of the silver image "by chemical treatment". Darrah identifies this process narrowly in the Boston area about 1870-1873, as applied to cartes de visite. Enlarged charcoal portraits were made in the same manner, and apparently also retouched by wax or pastel crayons.

This is the only reference found so far that mentions removal of the silver image after retouching, rather than weakening the image before retouching (leaving a dim image that is visible in our pictures). Darrah does not describe the chemistry, but various bleaches were available, some of which embrittled the paper; weak sulfuric acid is one such bleach. Our portrait showed serious paper crumbling, more than is usual with old photographs, which could have been the consequence of image removal, or just inferior paper. Different practitioners are known to have used many process variations.

A book by Barhydt, reference [19], published in 1892, is the only book solely devoted to crayon prints that this writer has encountered. It was found in the rare book section of the Library of the George Eastman House. Unfortunately the book is not informative about the various photographic processes. It describes the use of 'Conte crayons', which are still sold in artists' supply stores; they have been manufactured for two hundred years. They have a square cross-section and are hard and 'chalky', rather than waxy like our present-day children's crayons.

The Arizona Paper and Photograph Conservation Group held a symposium on December 2, 1989, at the Center for Creative Photography at The University of Arizona. The guest speaker was James Reilly, Director of the Image Permanence Institute in Rochester, New York. One topic was crayon prints. From the discussion concerning similar specimens, the specimen in our analytical study was definitely identified as a crayon print, and our analytical conclusions were essentially correct. We still would like more details of the photographic process; no doubt there were many variations among individual practitioners. But crayon prints evidently had considerable vogue.

The excellent book by Reilly [122 page 6] mentions crayon prints explicitly but does not elucidate the photographic process beyond mentioning the use of both POP and DOP processes. His book was published several years after our unpublished SEM analysis was performed.

Other Speculations

The presence of the other elements leads to some interesting speculations. With the exception of trace chlorine in sample #1, the remaining elements are not associated with silver bromide systems, and it seems likely that they may be constituents of tinting pigments or paper fillers. The following list of pigments containing these elements was compiled from tables of pigment compositions *. It is interesting to note some common pigment elements that were not detected, such as titanium, zinc, cadmium, mercury, copper, cobalt, sodium, arsenic, and manganese.
* Handbook of Chemistry and Physics, 61st Edition 1980-81, pages F85-86, CRC Press, Boca Raton, FL 33431.

Lead-containing pigments:
PbO - yellow litharge
PbSO4, PbCO3, Pb(OH)2 - white lead
Pb3O4 - red lead
Pb3(SbO4)2 - Naples yellow
Calcium-containing pigments:
CaSO4 - white gypsum
CaCO3 - white chalk
Iron-containing pigments:
Fe2O3 - red or yellow ochre or burnt sienna
Fe4[Fe(CN)6]3 - Prussian blue
Barium-containing pigments:
BaSO4 - white baryta
BaCO3 - white
Antimony-containing pigments:
Sb2S3 - vermillion
Sb2O3 - white
Pb3(SbO4)2 - Naples yellow
Sulfur-containing pigments:
BaSO4 - baryta white
Sb2S3 - vermillion
Silicon-containing pigments:
SiO2 - sand or diatomaceous earth.
The principal colors in the portrait are blue in the eyes and in the background wash, red lips and cheeks, and white tracery in the blouse. Yellow or green are not apparent.

The color of pigments depends not only on their chemical composition but also on their crystal structure, hydration, and on trace impurities. The electron microprobe could not detect hydrogen, carbon, oxygen, or nitrogen, which eliminates information on oxidation states, water of hydration, and organics.

With these caveats, the following compounds are possible but cannot be confirmed by the instrument used in this analysis:
PbO - yellow litharge
PbCO3, Pb(OH)2 - white lead
Pb3O4 - red lead
CaCO3 - chalk
Fe2O3 - red or yellow ochre
Fe4[Fe(CN)6]3 - Prussian blue
Sb2O3 - white
BaCO3 - white
SiO2 - sand
The following materials were not present, within detection limits:
Clay or kaolin (no aluminum found).
Talc (no magnesium found).
No Ultramarine pigment (no sodium or aluminum found).
The following compounds, all containing sulfur, may be present, depending on how the available sulfur is allocated (since we have no valence or bonding information):
PbSO4 - white lead
Pb3(SbO4)2 - Naples yellow
CaSO4 - gypsum
Sb2S3 -vermillion
BaSO4 - baryta
AgS - silver sulfide
Sample #1 showed antimony but no lead, while sample #2 showed lead and no antimony. It may be that two red pigments were used: Sb2S3 for vermillion and Pb3O4 for red. When we selected the loose flakes for analysis, their locations in the image were unfortunately not precisely noted. The cheek coloring appears to be a slightly different shade of red than the lips, so one may have been lead and the other antimony. We were originally more interested in the silver question than in identifying pigments.

We have indulged in these speculations to show some of the possibilities of non-destructive x-ray fluorescence analysis. It should be emphasized that a thorough analytical treatment would have required additional sampling, and compound information from other techniques such as infrared spectrophotometry.