JAIC 1978, Volume 18, Number 1, Article 4 (pp. 19 to 32)
JAIC online
Journal of the American Institute for Conservation
JAIC 1978, Volume 18, Number 1, Article 4 (pp. 19 to 32)

PIGMENT ANALYSIS OF EARLY AMERICAN WATERCOLORS AND FRAKTUR

Janice H. Carlson, & John Krill



3 PIGMENT ANALYSIS

ELEMENTAL ANALYSIS by X-ray fluorescence was used in this study of painting materials. The technique, previously used in the determination of the chemical composition of silver, brass, pewter and glass,13, 14 was modified slightly for pigment analysis so that the 55Fe, 241Am, and 109Cd sources are successively employed to obtain data in three different adjacent portions of the 0–40 keV spectrum. Potassium, calcium, titanium, and vanadium are detected by a 100 second irradiation on the 55Fe source; silver, tin, antimony, barium and others with a 100 second irradiation on the 241Am source, and copper, iron, zinc, lead, arsenic, mercury, chromium, and others of particular interest with regard to pigment analysis with a 300 second irradiation on the 109Cd source. Whatman No. 541 filter paper is used to obtain a baseline; x-ray fluorescence data from this paper are subtracted by the computer from the pigment data line. In addition, a clear unpigmented portion of the painting is analyzed when possible. In this way, any elements which may be in the paper can be manually subtracted out. Peak locations are determined by using a reference glass standard which contains thirty different elements. Count data and graphical plot of each pigmented area are obtained, then interpreted.

While pigment samples could have been removed from the fraktur and watercolors being studied, we chose not to do so. Rather, our efforts were directed toward demonstrating the utility for pigment analysis of a technique already available in our laboratory. Without the removal of any sample, even the most minute, x-ray fluorescence analysis, although not providing a complete description of the pigment, frequently provided sufficient information on elemental composition to permit reliable conclusions as to the identity of pigments. Further, the rapidity of the technique permitted the analysis of more than 100 pigmented areas in only a few days time.

X-ray fluorescence analysis does have certain limitations. Since the technique is not sensitive to elements below atomic number 19, pigments which are metal acetates, silicates, aluminates, etc. cannot be fully characterized. Similarly, organic dyes are not detected except by the absence of metal cations. If the pigmented area is small, the elements present in neighboring areas may obscure those in the area of interest. Finally, analytical data obtained on colors made of two or more pigments, e.g. Prussian blue mixed with an organic yellow to form green, may be subject to misinterpretation.

The first step in attempting to resolve the controversy surrounding the source of fraktur painters' materials was to examine the pigments used by watercolor painters in the period 1750–1850. To this end, the pigments in three watercolor paint boxes in the Winterthur collection, and those of a group of non-fraktur American watercolors from the same time period were analyzed.

Although the first box, of English origin, is itself authentic, dating from the late eighteenth century, analysis of the pigments showed that several of them were not available until sometime later. A yellow pigment, for example, contained both lead and chromium indicating it was chrome yellow (PbCrO4), a pigment which did not become available on a commercial scale until 1818. A second paint box (England, early nineteenth century) contained about forty watercolor cakes, rather than powdered pigments. The elemental compositions of these cakes are summarized in Table I. In a few instances in this Table as well as Table II, elements expected on the basis of the label identification are absent; in other cases, additional unexpected elements are present. More study on these is desirable but was not undertaken as part of this project. For the most part, the elements found in the pigment cakes and the pigment identifications based upon them are consistent with their attributed dates.

TABLE I ELEMENTS FOUND—PAINT BOX (68.1828) REEVES & INWOOD—ENGLISH, 19TH CENTURY

TABLE II ELEMENTS FOUND—PAINT BOX (65.1303) G. C. OSBORNE, PHILADELPHIA—AMERICAN C.1826

The third paint box examined was made by Osborne and Company of Philadelphia in 1826. Although the pigment cakes are fewer in number than in the second English paint box, one should not assume that American watercolorists had a limited selection of pigments available to them.

Note among the pigments in the Osborne box (Table II) the presence of vermilion, Prussian blue, chrome yellow, several iron containing umbers and siennas, and others.

Data from the analysis of the pigments of eleven non-fraktur watercolors, summarized in Table III and discussed below suggests, however, that even though large numbers of pigments were available, only a few were actually used to any extent.

TABLE III PIGMENTS INDICATED ON ELEVEN WATERCOLORS AMERICAN, C.1700–1850

  1. Red. All eleven watercolors had red areas, eight of which contained appreciable amounts of mercury. The presence of mercury indicates that the pigment used was vermilion, a red mercuric sulfide, a pigment in use since classical times, and manufactured in Europe since the early medieval period. An iron oxide red—probably red ocher, another ancient pigment—appeared in two of the eleven watercolors. A third red pigment, which visually appeared quite different from the others, was thought to be organic, probably a lake, since no x-ray fluorescence peaks attributed to red inorganic pigments were found.
  2. Blue. Prussian blue (iron ferricyanide), well known as a pigment in Europe by 1750, was used on six of the eight watercolors with blue areas. Azurite or blue verditer, a basic copper carbonate, appeared on one.
  3. Green. Verdigris, a basic copper acetate, and green earth, an iron-containing hydrosilicate, are equally represented among the watercolors tested—four each. Scheele's green or possibly Paris green, both characterized by the presence of copper and arsenic, appears on one watercolor.
  4. Yellow. A wider variety of yellow pigments was found. Chrome yellow, or lead chromate, not commercially available until 1818, appears on a watercolor with an attributed date of 1810–1825. The presence of chrome yellow places this watercolor at or near the very end of that period. An organic material, possibly gamboge or another yellow pigment, appears on two paintings, and orpiment on one. Yellow ocher, present on three paintings, seems to have been the yellow pigment used with greatest frequency.
  5. Brown and Black. The brown areas on half of the brown and black group were either iron containing pigments or iron gall ink. Differentiation between the various iron ochers, sienna, and inks is not possible by x-ray fluorescence. In one case, the iron was accompanied by arsenic, suggesting the use of an obscure pigment called terra di Siena which is not listed by the usual reference sources.15, 16 A similar pigment containing both iron and arsenic was found among the English pigment cakes. Bone black, indicated by the presence of a large calcium peak and the absence of any other peaks, was found on two of the paintings and carbon black on one.
  6. White. A distinctive feature of many of the watercolors was the ubiquitous presence of lead. Seven of the eleven watercolors contained lead—in white areas, in paler hues, and as an opacifier in the effort to achieve an effect more similar to that of oil painting.

How do the pigments of fraktur compare with those found on early American watercolors? The findings on sixteen fraktur are summarized in Table IV and discussed below.

TABLE IV PIGMENTS INDICATED ON SIXTEEN FRAKTUR AMERICAN, C.1780–1850

  1. Red. Fifteen fraktur had red pigmented areas which contained mercury, indicating the use of vermilion. The other red pigment found was red ocher. Both of these pigments were found on the non-fraktur watercolors.
  2. Blue. Prussian blue, the blue pigment of choice of watercolorists, was also widely used on the fraktur tested, appearing on nine. Smalt, which appears on one fraktur, is indicated by the presence of both cobalt and arsenic.
  3. Green. A copper green, in most cases probably verdigris, a basic copper acetate appears on ten fraktur. The fact that many of the green areas have turned quite brown indicates the use of verdigris rather than copper resinate in those areas, since verdigris readily decomposes in the presence of water and heat to form a dark residue, cupric oxide, while copper resinate remains somewhat protected by the resinous medium. Verdigris and the other green pigment occasionally used, green earth, have both been known and in use since antiquity.
  4. Yellow. Of the fifteen fraktur with yellow areas, the majority, ten, appeared to contain an organic material as a colorant. The supposition is that gamboge, an organic resin in wide use at the time, was the pigment used. Other possibilities include Indian yellow, annotto and saffron. Since x-ray fluorescence is not sensitive to organic materials, the pigment's exact identity is not known. Yellow ocher, another iron oxide, and orpiment, both widely used since classical times, are also represented. Of greatest interest is the presence of chrome yellow or lead chromate in three of the fraktur. All three fraktur which bear this pigment postdate the commercial appearance of chrome yellow in 1818, but not by much. This evidence strongly suggests that fraktur painters had ready access to new pigments soon after their introduction.
  5. Brown and Black. Since fraktur are frequently pen and ink drawings painted in with other colors, many of the brown and black areas are of the same iron gall ink used to outline the figures. A homemade iron gall ink was in common use by all members of society, not just Pennsylvania Germans. Other black and brown areas which differ in appearance from the ink areas and also iron-containing pigments are not differentiable by x-ray fluorescence from the iron inks.
  6. White. Lead white seems to have been used sparingly by fraktur artists. A lead-containing pigment was occasionally found mixed with a red pigment, probably vermilion, to make a pink color, or mixed with vermilion or Prussian blue to achieve an opaque effect. However, the presence of large quantities of lead white in all parts of a watercolor seems to be a technique used primarily by non-fraktur watercolorists. The fraktur artists were calligraphers who decorated their often functional documents with ornamental drawings, whereas for the non-fraktur watercolor artists, the emphasis lay in painterly techniques.

For further information on the pigments used, we turned to two fraktur painters boxes in the collection of the Mercer Museum17; see Figure 4. A blue material from one bottle was Prussian blue. Among the pigments found in the second box (Table V) were white lead, two samples of Prussian blue, chrome yellow, vermilion, and terra di Siena. The last color was previously found in one watercolor and the wooden English paint box. Its presence in one of these fraktur painter's boxes is particularly significant since it ties together both fraktur and non-fraktur watercolor paintings with a known commercially available but perhaps little used pigment.

Fig. 4. Fraktur Painter's Box (No. 25370), Courtesy of the Mercer Museum of the Bucks County Historical Society

TABLE V PIGMENTS FROM FRAKTUR PAINTER'S BOX MERCER MUSEUM NO. 25370

These data from the analysis of pigments on sixteen fraktur, together with those from the fraktur painters' bottles, refutes the idea that fraktur pigments came only from local farmlands and gardens. The confusion seems to have arisen from Henry Mercer's statement that the artist's paint bottles contained “home-mixed inks and paints”. Home-mixed—yes. The commercially bought powder or cake pigments were undoubtedly mixed with a medium or binder. These pigments were not, however, home-made, that is extracted from local plants and berries.


Copyright 1978 American Institute of Historic and Artistic Works