X-Ray Fluorescence as a Method of Characterizing Inorganic Pigment Patterns in the Work of Julian Onderdonk.

doi:10.21203/rs.3.rs-1340147/v1

This study utilized portable X-Ray fluorescence to analyze pigment patterns in 33 paintings by Julian Onderdonk, a 19th-20th century Texas impressionist. This analysis led to the identification of distinctive pigment preferences for Onderdonk at different periods of his career. Using the pigment preference patterns identified in the paintings that were dated by the artist, undated works were analyzed and assigned to different periods in the artists career based on their pigment patterns. This study represents a non-destructive method for organizing the artist’s work without solely relying on stylistic changes.

X-Ray Fluorescence

Art

Pigment Analysis

Non-Destructive Testing

Paint

Robert Julian Onderdonk (1882-1922), commonly known as Julian Onderdonk, was an American Impressionist painter from San Antonio, Texas. He is best known for his vibrant, sprawling depictions of bluebonnet-filled Texas landscapes, and his paintings have been showcased in museums and art galleries around the United States, even adorning the White House walls during the presidency of George W. Bush from the years 2000-2008 (Yoo 2015, Kahn 2014).

Originally from Texas, Onderdonk studied art at the Art Students League in New York City from 1901-1909, and while he sold his work the best he could, like many aspiring artists he struggled financially. In 1904, Onderdonk began a business relationship with someone known only as Chas. Tunison, (likely either Charles or Chase), who paid Onderdonk to paint and then sold the paintings on his own. The paintings that Onderdonk did for Tunison were signed under a pseudonym of Chas. and Chase Turner. The Turner paintings are indistinguishable from the work that Julian Onderdonk signed with his own name; often they are of the exact same subjects and scenes. By having him use a pseudonym and selling the work without his presence, Tunison was able to keep Onderdonk reliant on him without any alternative source of income for at least four years. By 1908, Onderdonk had either completely eliminated or at least severely limited his relationship with Tunison and began signing his own name to his work once again (Baker 2014:143).

The revelation that Chas. Or Chase Turner was a pseudonym for Onderdonk was made by Texas art historian James Baker, who published his findings in the 2014 book Julian Onderdonk in New York: The Lost Years – The Lost Paintings. The paintings analyzed in this study cover the three major periods of Onderdonk’s work, ranging from his pre-New York phase in 1899, through his time in New York from 1901 to 1909, followed by his time back in Texas from 1909 up to the very last painting he completed before his death in 1922.

Baker possessed or had access to 33 paintings signed both as Onderdonk and Turner, ten of which were undated and two of which had only tentative dates based on associations with similar, dated works. Analysis of Onderdonk’s signature revealed a shift in style between 1906 and 1908, allowing some of the undated works to be grouped to a general date range (Baker 2014: 127-129). To further narrow down the dates, this study utilized portable X-Ray fluorescence (pXRF) analysis to attempt to identify pigment markers that can further refine these groupings, and to develop pigment patterns that are unique to certain periods of Onderdonk’s career. By then comparing the pigment patterns in these refined groups to the undated works, date ranges can be assigned with greater certainty than stylistic comparisons.

For each painting analyzed, several small areas were chosen to examine based on pigment hue, apparent homogeneity of the paint, and the thickness of the paint. Data was obtained with a Bruker Tracer III-SD pXRF detector, and analyzed in Bruker’s S1pXRF software. The spectra for the different points were compared to one another in order to attempt to isolate the constituent elemental building blocks responsible for a particular hue. In some cases, a distinct pigment color was determined, and in others, the colors were clearly a blend of two or more colors. Though some of the pigments analyzed remain ambiguous, for the most part this method of analysis was successful.

There are two major drawbacks to using pXRF to analyze pigments. The first is that the Bruker pXRF instrument is unable to detect organic elements due to the low fluorescence energy levels of low Z-number elements. This means that the analyses presented here are likely incomplete. A number of organic based pigments exist and are frequently used, but given the inability of the instrument to detect them, they were necessarily ignored for this study. In some cases, it is noted that due to a lack of otherwise compelling data, organic pigments are the most likely source of a color, but this is inference based on a lack of evidence.

The second drawback is that fundamentally, XRF analysis is unable to detect chemical structure. The elemental composition is read, but the chemical bonds between the elements are not (Cesareo et al 2008, 206, McGlinchey 2012, 131). Many elements are present in a wide variety of pigments. Iron (Fe) for example can be found in hematite red, orange and yellow ochre, and even Prussian blue (Eastaugh et. al. 2004, 112, 144). When each pigment point is analyzed, the hue of the point must be taken into account when attempting to find the best chemical fit for the elements present. Below is a table summarizing the pigments identified in this study with their chemical formula and earliest known use.

Pigment name	Chemical Formula	Date
Barium white	BaSO₄	Used in Antiquity, manufactured from 1830CE in France
Titanium white	TiO₂	First produced in 1821CE, but not available to artists until the late 1920s CE
Lead white	Pb₃(CO₃)₂(OH)₂	At least as old as 1500BCE in Egypt and Greece
Zinc white	ZnO	First produced as early as 1782CE, mass produced by 1835CE
Realgar red	As₄S₄	Used in antiquity
Hematite (or ochre)	Fe₂O₃	Used since prehistoric times
Red lead (minium)	Pb₃O₄	Used since the Roman empire
Vermillion red	HgS	Used in antiquity
Lemon yellow	BaCrO₄ or SrCrO₄	Discovered in 1809CE
Cadmium yellow	CdS	Discovered in 1817CE
Chrome yellow	PbCrO₄	Discovered in 1809CE
Zinc yellow	ZnCrO₄	Discovered in 1809CE
Chromium oxide green	Cr₂O₃	Discovered in 1809CE
Cobalt green	Co – doped zinc oxide	Discovered in 1809CE
Verdigris green	Cu(C₂H₃O₂)₂ • ₂Cu(OH)₂	Used in antiquity
Azurite blue	Cu₃(CO₃)₂(OH)₂	Known since the 15th century, manufactured artificially since the 18th century
Cerulean blue	CoO • SnO₂	Introduced in 1821, reintroduced in 1860
Cobalt blue	Co – doped alumina	Discovered in 1775, manufactured from 1804 for artists
Egyptian blue	CaCuSi₄O₁₀	Used in antiquity
Han blue	BaCuSi₂O₆	Discovered in 1859
Prussian blue	(Fe₄[Fe(CN)₆]₃)	Known since 1704, manufactured from 1870 for artists
Smalt	Co – doped Si glass	Known since mid-16th century
Cobalt Violet	Co₃(PO₄)₂ or Co₃(AsO₄)₂	Discovered in 1859

Table 1: Pigment names and chemical formulas (From Artioli 2010, Roy 1986, and Wehlte 1975).

2.1 Sample Painting Analysis

Elemental data points were collected across all 33 paintings, then collated and compared. An example of how the analysis was carried out across the assemblage can be seen here with the 1922 painting Dawn in the Hills (Figure 1). Apart from being a particularly beautiful painting in its own right, Dawn in the Hills is highly treasured due to the fact that it is the last painting that Onderdonk ever painted (Rudolph 2008, 40).

A total of seven points were analyzed for this painting. Points 1 and 2 are the whites of the sky, points 3 and 4 are blues, point 5 is the green of the tree to the left of the painting, point 6 is the reddish background in the lower left-hand corner, and point 7 is the signature. This painting was framed and unable to be removed from the frame, but for those that were unframed or able to be removed from a frame, a portion of non-painted canvas was analyzed to determine the elements present in the primer. After each point was tested, they were analyzed by color grouping.

2.1.1 White

The elemental composition of the white of the clouds in points 1 and 2 are predominantly zinc and lead, and therefore are most likely to be a zinc white and a lead white. Both zinc and lead peaks are consistent across all peaks, and so it stands to reason that the painting was primed with both.

2.1.2 Blue

Both blues of points 3 and 4 are clearly cobalt based when compared to the white spectrum of point 1. As with the rest of Onderdonk’s paintings in the 1920s, this cobalt based blue does not have the indicative tin or aluminum peaks that would point toward cerulean blue or cobalt blue. Nonetheless, this cobalt signature is consistent with the other blues in Onderdonk’s late period work.

2.1.3 Green

The green of point 5 is a chromium based green, almost certainly chromium oxide (emerald green).

2.1.4 Red and Signature

The area behind the signature is a reddish orange, which appears to be iron based, indicating hematite. In point 6, there is also a trace of mercury, indicating that the red highlights are vermillion (HgS) instead of iron. Point 7, the signature, has a slightly higher peak for chromium, likely indicating that the red and green were mixed to make a darker color.

For Dawn in the Hills, as the spectra were analyzed, the results were added to a spreadsheet for comparison across the collection. As the results continued to build, patterns and trends started to become clear.

Table 3 lists each painting with a known date and the characteristic elements for each hue analyzed, with the exception of Bluebonnets in San Antonio; this painting had titanium in each point analyzed, indicating that it was primed with titanium white, a paint that was not available until after Onderdonk’s death. Additionally, Milkweed in Bandera was omitted due to an inability to separate out conclusive elemental data for the different pigments. Based on this table, Onderdonk’s preference for zinc white over lead white grew, and by 1914 it seems he had phased the lead white out of use almost entirely until he re-introduced it in 1921. He seems to have eliminated mercuric red pigment by 1908, though it does show up again in the 1922 Dawn in the Hills. In 1909 there was a shift away from iron-based blues to cobalt-based blues, along with a shift from zinc-based yellows to cadmium-based yellows.

Painting	Year	Primer	White	Red	Yellow	Green	Blue	Brown
Sunset November 1899	1899	Pb	Pb	Fe	Zn	Cu	Org/Cu
Sand Dunes	1901	Pb	Pb	Fe/As	Zn / Ba	Cu	Fe
Cock Fight Amid the Jacals	1905	Pb	Pb	Hg	Fe / Sr	Cr	Org	Fe
Portrait of G. Beddle Moore	1905	Pb	Pb	Hg	Zn / Ba			Mn/Fe
Harbor Scene	1906	Pb	Pb	Hg			Fe	Fe
October Sullivan Co. NY	1908	Pb	Pb	Fe	Zn	Cr	Org /Fe
A June Morning	1909	Pb	Pb	Fe	Zn / Ba	Cr	Fe
Family at Cards	1909	Pb	Pb	Fe	Zn / Cd		Co
Bluebonnets and Cacti	1910	Zn / Pb	Zn / Pb	Fe	Sr / Cd	Cr	Co / Fe
Bluebonnet Field	1912	Pb	Pb	Fe	Cd	Cr	Co / Sn
A Pool on the Guadeloupe	1913	Pb	Zn / Pb	Fe	Zn	Cr	Co / Sn
Afternoon Back of Laurel Heights	1913	Pb	Zn / Pb	Fe	Zn / Cd	Cr	Co
Evening in the Bluebonnets	1914	Zn	Zn	Pb / Fe	Cd	Cr	Co / Sn
Texas Mountains on Williams Ranch	1915	Zn	Zn / Ti	Pb / Fe	Cd		Co / Sn
Williams Ranch, 20 West of Kerrville TX, near Junction	1915	Zn	Zn	Pb / Fe	Cd	Cr	Co
Late Afternoon Sunlight on the Bluebonnets	1916	Zn	Zn	Fe	Cd	Cr	Co / Sn
Bluebonnets at Sunset San Antonio	1919-20	Zn	Zn	Fe	Ba / Sr		Co / Sn
A White Road at Late Afternoon	1921	Zn	Zn	Pb / Fe	Ba / Sr	Cr	Co
Miles and Miles of Bluebonnets	1921	Zn	Pb	Fe	Sr	Cr	Co

Table 2:Dated paintings and their primary elemental constituents for each color.

Using this table as a stepping off point for spectral comparisons between paintings, the elemental composition of the dated paintings were compared to the undated paintings to see which paintings fit with which date range the best. Several paintings illustrate shortcomings in the date ranges derived from the dated paintings; At the Edge of the Forest fits best with the 1909 range of pigments, but the inclusion of mercury hints that Onderdonk continued to use mercuric red throughout his entire time in New York instead of stopping in 1906. Likewise, Seascape V, signed Chas. Turner, fits best with the dates of 1905-1908, and if this is correct, the supposition that zinc whites were not introduced until 1909 must be incorrect.

Painting	Primer	White	Red	Yellow	Green	Blue	Brown
Pastoral with Lady in Red	Pb	Pb	Hg / Fe	Ba	Cr	Fe	Mn / Fe
Seascape V	Pb / Zn	Pb / Zn	Hg /Fe	Ba	Cr	Org
Landscape with Cacti	Zn	Zn	Pb	Cd	Cr	Co
A Man Fishing From a Boat	Pb	Pb	Fe / Hg	Zn/Ba	Cr	Cu
At the Edge of the Forest	Pb	Pb	Hg / Fe	Cd/ Zn	Cr
Central Park	Pb	Pb	Fe	Cd	Cr		Fe
Fall Landscape #2	Pb	Pb	Hg	Ba/Sr/Zn	Cr	Cu
Large Bluebonnet Landscape	Zn	Pb	Fe / As	Cr/Sr	Cr	Cu / Co
Summer Afternoon	Pb	Zn / Ba	Fe		Cr	Org	Mn
Bluebonnet on a Texas Landscape	Zn	Pb	Fe	Cd	Cr	Co / Sn	Mn
Study - Lupine	Zn	Zn	Pb	Cd/ Ba Sr	Cr	Cu / Co

Table 3:Undated paintings and their primary elemental constituents for each color.

Despite these shortcomings, the compositions of several of the undated paintings match very well with the dated paintings. Pastoral with Lady in Red looks very much like Cock Fight Amid the Jacals and Portrait of G. Beddle Moore, and it is likely that it was painted around 1905. Table 4 lists the undated paintings with their most parsimonious compositional match to the dated works. To further substantiate the matches made via the elemental composition, the undated paintings were crosschecked with the signature analysis of selected paintings carried out by Baker (2014, 165-168). There were no conflicts between the assigned date ranges and the signature estimated dates, which, if nothing else is at least encouraging.

Painting	Assigned Date Range	Basis	Signature Estimated Date
Pastoral with Lady in Red	1905	Pb white, Hg red, Fe blue, Mn brown
Seascape V	1905-1908	Hg red, Org Blue, Ba yellow
Landscape with Cacti	1910-1915	Zn white, Pb red, Co blue
A Man Fishing From a Boat	1905	Zn/Ba yellow, Hg red, Pb white	1905
At the Edge of the Forest	1909*	Pb white, Cd yellow
Central Park	1909-1912	Pb white, Fe red, Cd yellow	Post 1906-1908
Fall Landscape #2	1899-1906	Pb white, Hg red, Zn yellow, Cu blue	1905-1908
Large Bluebonnet Landscape	1910	Pband Zn white, Fe red, Sr yellow, Co blue	Post 1906-1908
Summer Afternoon	1905	Org blue, Mn brown	Pre 1908
Bluebonnet on a Texas Landscape	1910-1913	Pb and Zn white, Fe red, Cd yellow, Co and Sn blue, Mn brown	1910
Study - Lupine	1915-1920	Zn white, Pb red, Ba and Sr yellow, Co blue

Table 4:Undated paintings and their assigned date ranges based on spectral similarities with dated paintings.

* The inclusion of Hg red could suggest an older date, but Seascape V supports a 1909 date.

Despite the inability to directly determine the chemical bonds of elements in a material using XRF, with enough data points and a moderately restricted pool of possible chemical formulas, it is entirely possible to establish the chemical makeup of cultural resources like art and archaeological artifacts. With the work of Julian Onderdonk, clear patterns of changing pigment preferences through time revealed themselves when each of the paintings was analyzed. It has been shown that undated works can be reasonably associated with the spectral signatures of dated works, allowing for a completely non-destructive way to categorize the undated works into the various phases of Onderdonk’s career. It has also been shown that by establishing these patterns, paintings incorrectly attributed to an artist can be discovered, as is most likely the case with Bluebonnets in San Antonio.

The aim of this study was to utilize a portable non-destructive analysis technique to expand the knowledge available about a collection of cultural heritage materials. The results of the study outlined here have proven to be very satisfactory in achieving this aim, though clearly refinements in instrumentation would be advantageous. Future studies incorporating the use of analytic methods capable of detecting organic materials and the chemical bonds between the elements detected would greatly enhance the breadth and depth of the results outlined in this study. However, given that the pXRF was easily transported to the various locations of the paintings and obtained excellent data sets without any adverse effects on these priceless works of art, this was overall a successful endeavor.

As Arsenic

Ba Barium

C Carbon

Cd Cadmium

Co Cobolt

Cr Chromium

Cu Copper

Fe Iron

Hg Mercury

O Oxygen

Pb Lead

S Sulfur

Sr Strontium

Ti Titanium

Zn Zinc

pXRF Portable X-Ray Fluorescence

Availability of data and materials

All of the spectra files obtained for this study from any or all of the 33 painting analyzed is available by request from the Author, either in .pdz or CSV format, or a .jpg screenshot.

Competing Interests

The author declares that they have no competing interests

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Authors Contributions

This is a sole-author manuscript. All research, analysis, and writing was done by CD.

Acknowledgments: This study was made possible by the generous access to several collections, including that of James and Kimel Baker, the Witte Museum and the Villa Finale Museum & Gardens of San Antonio, TX. Additionally, the Bakers facilitated access to this museum, and were vital to the success of project. The use of the pXRF unit was made possible by Texas A&M University’s Center for Maritime Archaeology and Conservation.

Artioli, Gilberto. 2010. Scientific Methods and Cultural Heritage: An Introduction to the Application of Materials Science to Archaeometry and Conservation Science. Oxford University Press. Oxford, UK.
Baker, James. 2014. Julian Onderdonk in New York: The Lost Years, The Lost Paintings. Texas State Historical Association. Denton, TX.
Cesareo, R., Ridolfi, S., Marabelli, M., Castellano, A., Buccolieri, G., Donativi, M., Gigante, G., Brunetti, A., Medina, M.. 2008. Portable Systems for Energy Dispersive X-Ray Fluorescence Analysis of Works of Art. In. Portable X-ray Fluorescence Spectrometry: Capabilities for In Situ Analysis. 2008. Ed. Potts, P, and West, M. Royal Society of Chemistry Publishing. Cambridge, UK.
Eastaugh, N., Walsh, V., Chaplin, T., and Siddal, R. 2004. Pigment Compendium: Optical Microscopy of Historical Pigments. Elsevier Butterworth-Heinemann. Burlington, MA.
Kahn, E. 2014. Historians’ Book Reveals a Texan’s Noms de Brush. New York Times. 7 March 2014. Section C, Page 34.
McGlinchey, C. 2012. Handheld XRF for the Examination of Paintings: Proper Use and Limitations. In: Studies in Archaeological Sciences: Handheld XRF for Art and Archaeology. Shugar and Mass, Ed. Leuven University Press. Leuvan, Belgium.
Rudolph, W. 2008. Julian Onderdonk: American Impressionist. Yale University Press. New Haven Ct.
Wehlte, K. 1975. The Materials and Techniques of Painting. Translated by Ursus Dix. Van Nostrand Reinhold Co, New York, NY.
Yoo, A. 2015. Breathtaking Impressionist Paintings of Bluebonnets by Julian Onderdonk. http://www.mymodernmet.com/profiles/blogs/julian-onderdonk-impressionist-paintings-bluebonnets

No competing interests reported.

X-Ray Fluorescence as a Method of Characterizing Inorganic Pigment Patterns in the Work of Julian Onderdonk.

Status:

Journal Publication

Version 1

Abstract

Figures

1. Introduction

2. Methods

3. Results

4. Discussion And Conclusion

Abbreviations

Declarations

References

Additional Declarations

Status:

Journal Publication

Version 1