Mineralogy and History of The Borghese-Windsor Cabinet

At the annual Gemological Institute of America research conference held in Carlsbad, CA this year (2018), I presented a recent collaborative study with Natural History Museum of LA, the Getty Museum, and Caltech.

The Getty Museum in Los Angeles recently acquired the Borghese-Windsor Cabinet (shown below), a piece of furniture extensively decorated with agate, lapis lazuli, and other stones. The cabinet is thought to have been built around 1620 for Camillo Borghese (later Pope Paul V). The Sixtus Cabinet, built around 1585 for Pope Sixtus V (born Felice Peretti di Montalto), is of similar design to the Borghese-Windsor and also ornately decorated with gemstones. Although there are similarities in gemstones between the two cabinets, they vary in their agate content. It was traditionally thought that all agate gemstones acquired during the sixteenth and seventeenth centuries were sourced from the Nahe River Valley near Idar-Oberstein, Germany. While Brazilian agate began to be imported into Germany by the mid-1800s, it is possible that some was imported in the eighteenth century or earlier. A primary research goal was to determine if the agates in the Borghese-Windsor Cabinet are of a single origin, or if they have more than one geologic provenance.

Borghese-Windsor Cabinet, now on display at the Getty Center in Los Angeles, CA. Photo from the Getty Museum.

Agates are made of silica (silicon dioxide), mostly as the mineral quartz, but also as metastable moganite. Both quartz and moganite will crystallize together as the agate forms, but moganite is not stable at the Earth’s surface and will convert to quartz over tens of millions of years (Heaney, 1995; Gíslason et al., 1997; Moxon and Rios, 2004). Thus, older agate contains less moganite. Agate from Idar-Oberstein area is Permian in age (around 280 million years old), while agate from the Brazilian state of Rio Grande do Sul generally formed during the Cretaceous (around 120 million years ago). It is thought that Rio Grande do Sul would have been a primary source of material exported to Europe because it is one of Brazil’s oldest and largest agate producers, and so we used these agates as our initial test material. Since Cretaceous agate from Brazil is many millions of years younger than Permian agate from Germany, the quartz to moganite ratios between the two localities should be quite different.

The agate gemstones of the Borghese-Windsor Cabinet could not be removed for detailed Raman spectroscopy mineralogical mapping experiments. Because of this, we first analyzed multiple agate specimens from the collections of the Natural History Museum of Los Angeles (NHMLA) and the Smithsonian Institution National Museum of Natural History (NMNH) using three different techniques: Raman mapping, X-ray fluorescence (XRF) mapping, and visible-shortwave infrared hyperspectral imaging. Raman spectroscopy provides an easy method to distinguish the relative quartz to moganite ratios, and XRF analysis provides a measure of bulk geochemistry in agates. Maps have advantages over line-scans and point analysis in that they give a better representation of the mineral content, can be used to exclude trace mineral impurities, and yield better counting statistics and averaging. Hyperspectral imaging provides a range of optical data from IR through visible wavelengths.

We performed Raman mapping and XRF mapping at NHMLA on many agates of known provenance that could later be compared to gemstones in the Borghese-Windsor Cabinet. When examining the cryptocrystalline parts of agate from comparative collections, Brazilian agates (as seen in figure below, from the NHMLA collection) had 8% or higher moganite concentration, whereas the Idar-Oberstein agate (on loan from NMNH) had less than 2% moganite.

Example of Brazil agate analysis (NHMLA collection). Left: Hyperspectral image showing the intensity of the H2O abosrption band. Right: Regular photograph of the agate, with Raman map showing the distribution of quartz(white)and moganite(black,darkest areas are approximately 8% moganite). All the Brazilian agate from Rio Grande Du Sol had this characteristic pattern of high moganite content in the cherty out-bands. The inner, larger quartz crystal parts, had little to no moganite.

Example of German agate analysis (NMNH collection). Left: Hyperspectral image showing the intensity of the H2O abosrption band. Right: Regular photograph of the agate. Raman maps(notshown) indicated the presence of < 2% moganite at the highest concentration. All the German agates we analyzed had little to no moganite in the cherty outer bands. The inner, larger quartz crystal parts, also had little to no moganite.

No intermediate moganite percentages were observed for the specimens we examined. Well-crystallized areas and microcrystalline areas within the same agate had little to no moganite. The moganite distribution in the agate is heterogeneous (see figures above), likely due to different growth stages and changing geological conditions during agate formation. Using the Raman maps, we were able to isolate the areas that contained moganite + quartz and measure the ratios in those specific bands. This narrow-band approach to determining quartz to moganite ratio, when compared to broad-brand and whole-sample approaches, was shown to be more reproducible in distinguishing Brazilian from German agates. Because moganite is isolated to select layers/bands within agate, care must be taken when evaluating quartz to moganite ratios for provenance or age analysis.

These same agates from the Brazilian and German localities (we analyzed a bunch) were then taken to CalTech (with help from Dr. Bethany Elhmann and Dr. Rebecca Greenberger) to collect hyperspectral imaging data (on a custom-built Headwall Photonics co-boresighted visible/near-infrared and shortwave infrared sensor). Imaging data were compared to the NHMLA laboratory Raman and XRF analyses, and correlation analysis of combined datasets from the three different experimental procedures allowed us to establish a unique characterization pattern for the different localities. At the Getty Museum (with the help of Getty conservator Arlen Heginbotham), we performed hyperspectral imaging of the entire cabinet and these analyses were compared to the data collected from known localities to attempt to determine provenance of the Borghese-Windsor agates. Because moganite is isolated to select layers/bands within agate, care must be taken when evaluating quartz:moganite ratios for provenance or age analysis.

Hyperspectral image showing the intensity of the H2O absorption band. Bright areas have more H2O and dark areas have less.

Color photograph of cabinet for comparison depicting the general area used for hyperspectral mapping. Photo from the Getty Museum.

In the end, we think there are two populations of agate. The first population seems to correlate well with our laboratory analyses of the agate from Idar-Oberstein. These are the dark patches where agate is located in the hyperspectral image of the cabinet shown above. The second population (the brighter agate areas in the hyperspectral image of the cabinet above) do not show the strong red colors seen from the laboratory studies of Brazilian agate, but they are brighter than what we believe could be from Germany.

So it is entirely possible that about half the agate on the chest is from Germany, and the other half from a different agate locality. Data analysis is still ongoing and we are also doing more analyses from agate localities from around the world. In addition, we are doing more work on getting a better understanding of moganite distribution in agates within the same locality.

Aaron Celestian is the Mineralogy Curator at the Natural History Museum of Los Angeles. He researches how minerals interact with their environments and with living things, and how those minerals can be used to solve problems like climate change, pollution, and disease.

All photos are my me unless otherwise noted in their caption. Getty images are open content, no permission required.

The above work was presented at the annual Gemological Institute of America meeting in October 2018. This conference proceeding will be published in Gems & Gemology in November 2018. Special thanks to the Smithsonian Institute for loaning specimens 103669–2, 106559, B5536, and C1316.