In 2014, Library of Congress Curator Carol Lynn Ward-Bamford reached out to the Library’s Preservation Research and Testing Division over concerns about possible deterioration among the Dayton C. Miller (DCM) collection of rare glass flutes made in the first half of the 19th century Paris by Claude Laurent. Using a microscope and a portable X-ray fluorescence spectrometer (XRF), Research Chemist Lynn Brostoff made the startling discovery that most of the DCM Laurent flutes are not made of “crystal glass,” i.e., leaded glass, as has been commonly assumed, but of an unstable type of alkali silicate glass called potash glass. This finding was surprising because Laurent was awarded a patent in 1806 for “flûte en cristal,” and there is no known mention of the maker selling glass instruments made of any other type of glass material.  The finding is significant because potash glass is well known to be prone to instability, and clearly is associated with deterioration that the two collaborators then documented with the aid of Dana Hemmenway of the Library’s Conservation Division.

This work initiated a detailed technical and historical study of the 20 DCM instruments, as well as other Laurent flutes and piccolos in worldwide collections. This led to the collaborative grant project described on this site, including a detailed technical study of Laurent’s oeuvre, including flutes by his apprentice Breton and a related instrument by Charly. Technical analysis has focused on non-invasive, accessible techniques: microscopy, visible ultraviolet (UV) fluorescence, pH measurement, non-invasive XRF, fiber optic reflectance spectroscopy (FORS), optical coherence tomography (OCT), and computed tomography (CT). Analysis has also included limited microsampling or surface sampling for examination by SEM/EDS and Raman spectroscopy. Partners at the Vitreous State Laboratory have contributed model glass studies based on replica glass with compositions modelled on those used by Laurent. Historical research has revealed new information about Laurent himself, and led to a new database of Laurent flutes worldwide. Overall, the technical and historical study has greatly enriched the story behind these remarkable musical instruments.

Highlights of findings

X-ray fluorescence (XRF) has allowed detailed compositional analysis of a good portion of extant Laurent flutes around the world, starting with the 20 DCM flutes, which greatly outnumber Laurent flutes in any other collections. As mentioned, Laurent’s original 1806 patent was awarded for leaded glass flutes, commonly known as “crystal glass.” In France, according to a mid-19th century source, the term “crystal” was reserved for glass containing lead oxide over about 25% in composition. On the other hand, Laurent’s typical potash glass formulation, as analyzed in this collection in atomic weight percent (at.wt.%), contains about 12–18 at. wt.% K, 2–3 at. wt.% Ca, and remaining elements other than Si less than about 0.2 at. wt.%. This potash glass appears to have been made as imitation “crystal,” similar to types being produced in Bohemia at that time. To date, only two out of 45 flutes by Laurent have proved to be leaded glass, including the beautiful DCM Madison flute. In addition, five flutes and piccolos in other collections have been found with one or more leaded glass joints. The remaining flutes examined have proved to be potash glass, so that it is not surprising that the bulk of Laurent’s output now shows signs of deterioration.

Deterioration in the glass instruments appears to be limited to those made of potash. This can be related not only to the composition, but to their specific history of environmental exposure and use. In general, it is well known that alkali silicate glasses are prone to deterioration by exposure to moisture, while leaded “crystal” glass is relatively stable as long as the lead content stays within a certain range. In addition, it has been observed that many of the potash instrument head joints are in worse condition relative to other joints in the same instrument. The head joints would have been in direct contact with the player’s mouth, face, breath and spittle, and thus exposed to increased levels of moisture, so that this makes sense. Laurent’s normal inclusion of extra joints for obtaining different pitches inevitably resulted in some joints being played less often, and these often appear in relatively good condition. Laurent flute DCM 475 is a perfect example: here, the long upper body joint of DCM 475 (third from top in picture) appears to have been an alternate joint and less played, and is visibly much clearer than other joints. This clarity indeed corresponds to better condition, while the quite opaque head joint area below the cork corresponds to advanced cracking and deterioration.

The use of UV-induced fluorescence has proved to be a reliable tool for distinguishing Laurent’s “crystal glass” flutes or flute joints from those made of potash glass. This is due to the inclusion of trace amounts of manganese oxide in the potash glass formulation, a technique that used for decolorizing glass since Venetian “cristallo” glass of the Renaissance era. The manganese (Mn) causes a characteristic green-yellow fluorescence under UVA illumination (356 nm), but only a dull pinkish fluorescence in the leaded glass flutes (unless trace Mn is also present). With UVC (254 nm) illumination, the composition can be confirmed due to the distinct icy-blue fluorescence emitted by high-leaded glass. Under UVC, the potash glass shows no visible fluorescence, even when there is trace lead (Pb).

Microscopy and OCT imaging have revealed various levels of deterioration among the DCM Laurent glass flutes, sometimes even among individual joints. These techniques have been used to document the shape and depth of visible symptoms of deterioration. OCT imaging furthermore allowed measurement of the glass alteration layer, which contains cracks and delamination. Liquid droplets have been observed with high magnification on the interior of a single joint of only one DCM flute, as well as on exterior and interior surfaces of some Laurent flutes in other collections. Surface accretions associated with deterioration have appeared to be minor among the plentiful debris and polishing residues typically seen on both interior and exterior surfaces of the DCM Laurent flutes, as confirmed by XRD and Raman spectroscopy.

SEM/EDS analysis of a microsample taken from a damaged extra foot joint of Laurent flute DCM 717 shows that a surprisingly uniform alteration layer in this area (adjacent to an existing hole). The deteriorated glass surface layer is about 30 µm thick, with cracks reaching through the layer roughly normal to the surface. EDS analysis of spots on the sample that start at the surface and reach through the alteration layer into the bulk glass show that potassium (K), as well as the trace amount of sodium (Na) present, are severely leached from the alteration layer.

A relatively new application of OCT was used in this study to dramatically document deterioration features beneath the glass surfaces of various flutes. 3D renderings are useful tools for non-invasively analyzing features below the glass surface that reflect the infrared light source. Such features in deteriorated glass include cracks, which can be visualized in terms of their shape and depth. As shown in the figure at the right, a light microscope image of the visibly undeteriorated surface of DCM 475 extra upper body joint (upper left) shows a series of parallel polishing marks across the surface, along with a variety of particulates that were identified mostly as debris and polishing residues.  The OCT 3D rendering of one area of this joint (upper right), viewed from underneath, confirms and documents this surface texture in a useful manner, confirming no evident deterioration. The 3D image also contains an artifact: a regular column-like structure extending into the glass.

The two figures beneath reveal a different story that describes significant deterioration on the lower body joint. Here, the light microscope image of the surface of DCM 475 lower body joint (below left) clearly shows extensive polygonal cracking, which is also visible as opacity to the naked eye. The 3D OCT image (below right) provides a stunning visualization of the cracking pattern as it extends into the glass, viewed from underneath as if we were inside the glass! The cracks are shown to extend to a depth of about 200 µm. Such 3D renderings literally expand our understanding of cracking that corresponds to what we see on the surface of historical alkali silicate glass, which should be considered before any treatment is proposed.

Average pH readings taken from flutes on both interior and exterior surfaces using pH colorpHast strips, a common laboratory method, ranged from 6.5 to 9.5. However, these results are confusing, since they often do not correlate to the visible condition. For example, pH on the severely degraded head joint of DCM 717 measured only slightly alkaline (pH 8.1-8.7), while the extra foot joint, which shows little visible deterioration, measured pH 8.3-9.5. However, OCT analysis of several flutes showed that alteration layers may vary in thickness from spot to spot. This likely explains results such as obtained on the severely degraded DCM 1680, where pH measurements from exterior surfaces of all joints averaged only 7.5 (about neutral), and interior surfaces averaged 8.6 (slightly alkaline), but with one interior spot measured pH 10.0 (very alkaline). This particular flute was acquired in 2018, and it is possible that it was cleaned prior to sale, and wiped down with some frequency in its former collection. These results demonstrate that pH measurement with paper strips on a limited number of spots does not reliably indicate the state of deterioration on any one object by itself. While findings of pH > 8 are likely warning signs of deterioration, in all cases, pH measurement should be combined with other tools for proper assessment of glass instability.

The consistency in production between Laurent flutes, as well as each joint, was dramatically revealed by  3D CT. Two Laurent flutes, DCM 1680 (1819) and DCM 1681 (1828), were scanned at the George Washington University Department of Radiology. The axial slices, shown in the video to the left, show inner and outer diameter of the flute at 1 mm increments. The diameter data was measured using a custom Matlab code by Dr. Stephanie Zaleski. Both flutes show nearly identical interior measurements, suggesting that Laurent’s flutes were likely produced in a systematic, standardized manner throughout his career. The similarity in measurements stresses the importance of the flutes having similar musical qualities.

Fiber-optic reflectance spectroscopy (FORS) was utilized to analyze all 20 DCM Laurent flutes, plus the related DCM Charly piccolo. FORS results acquired from the leaded glass instruments (DCM 378, DCM 1051 and Charly piccolo) do not exhibit spectral features associated with glass hydration, consistent with observations that they are in fine condition. On the other hand, FORS spectra of the potash glass instruments exhibit distinct features near 1410 and 1910 nm associated with glass hydration in keeping with the visible deterioration. For example, the head joint of DCM 475 exhibits dense polygonal microcracking and also shows relatively large hydration peaks, while the long upper body joint of DCM 475 exhibits minimal deterioration and shows correspondingly smaller hydration peaks.

decision tree

Analytical results obtained from the Laurent flute technical study have been used to construct a basic “decision tree,” which is meant as an aid to conservators and curators in the assessment of unstable glass. Due to lack of data for other types of 19th century vessel glass, the tree is designed for use with Laurent flutes or glass of similar composition, and is meant to be used with the Visual Vocabulary for deterioration. The decision tree progresses from simple tools that are inexpensive and accessible to more sophisticated tools.  It is important to emphasize that each tool has its own limitations, and that multiple tools can increase the user’s degree of confidence in the preservation recommendation. High priority objects should be assessed by a trained glass conservator.