The Sergio: An Exploration of the World’s Largest Carbonado – Part 1

Acknowledgement

All images are © The Trustees of the Natural History Museum, London unless otherwise stated.

Abstract

In 1895, the largest carbonado ever discovered was found in Bahia Province in Brazil by a field worker named Sergio Borges de Carvalhoo. Dubbed “The Sergio”, this carbonado was sold several times then broken up for use as an abrasive for drilling. It was thus lost to science, and it seemed that all that remained for interested researchers was an image of an engraving published in a contemporary description by Henri Moissan, and an unattributed photograph reproduced in a number of articles relating to carbonados. Then, two models of The Sergio dating from 1911 were discovered in the collection of the Natural History Museum, London, allowing considerably more information to be gained about its size and shape. These discoveries prompted us to revisit the story of The Sergio and uncover its history. Here, in Part One of our research journey, we provide a comprehensive description of the models and use them to gain a better understanding of The Sergio’s dimensions and appearance. In Part Two, we trace the astonishing life and times of The Sergio from its finding to its destruction, a journey that led us to a remarkable discovery in the mineral collections of the Muséum national d’histoire naturelle, Paris.

Introduction

In a recent article on black diamonds in this journal (Hansen and Rennie, 2022), we wrote about “The Sergio” carbonado found in the Bahia province, Brazil in 1895, and named after the field worker who extracted it from the river deposits. It is generally accepted to be the largest carbonado ever discovered. In the age of industrial expansion and infrastructure building around 1900, carbonados were the hardest grinding material available. Our findings revealed that this massive carbonado was broken up and sold for its value as an abrasive, although the details of when and where this was done were unclear.

Privately, we lamented the permanent loss of this marvellous stone. We felt that the picture of an engraving and the single photographic image found in our searches, and reproduced in our article, seemed not to do justice to its incredible size and appearance.

A surprise was in store for us, or more literally “in storage”, at the Natural History Museum, London (NHM). Whilst assisting a visitor who was researching models of diamonds, the first author discovered not one, but two attractive replica models of The Sergio! Catalogued under “models”, not diamonds or carbonados, these models had been overlooked in our previous searches for carbonados. Finding them gave us the opportunity to confirm some of the suppositions ventured in our recent article and expand our knowledge of the shape and history of the remarkable original specimen.

But first we needed to establish that the models did indeed depict The Sergio. We had identical photographs published in four places: the guide for the Louisiana Purchase Exposition held in St. Louis (Brazil, 1904), Kunz (1904), Furniss (1906; the photograph used in our earlier article), and Yawger (1907), who claimed to have the original photograph, taken in June 1895, in his possession. Unfortunately, Yawger’s published image is of poor quality, so we chose the one published by Furniss (1906) as the clearest photograph to begin our research. This image is shown in Figure 1, juxtaposed with a photograph of model BM.1985,MI9558, one of the two models of The Sergio we describe below. The similarity is unmistakeable.

Confident that the models did represent The Sergio, we set off on a detailed exploration of the shape and appearance of The Sergio and its history. The twists and very surprising turns in our journey took considerable time and we relate it in two parts. In this first part we describe the models in detail and begin our search into the history of The Sergio. In Part Two of our journey, we uncover the remarkable history of The Sergio from its likely origin, to its finding and to its unfortunate destruction.

The Sergio Models

The reason the NHM has two models of The Sergio is historical. In 1985, the mineral and gem specimens of the Museum of Practical Geology merged with the collections of the NHM, previously the British Museum (Natural History), and each possessed a model (see Note 1 for a history of the two museums). This is not unusual – other examples exist within the NHM collection where similar or counterpart specimens were acquired at about the same time from the same source and came together with the amalgamation of the two collections.

The two models are shown in Figures 2 and 3, side by side, in as close to the same orientation as could be managed, with the Museum of Practical Geology model, BM.1985,MI9558, on the left and the British Museum (Natural History) model, BM.1911,119, on the right. Figure 2 presents the “front” of the models, and Figure 3 shows the “reverse” side. Both models are the same size, measuring 8.8mm x 8.2mm x 6.2mm, and clearly, they are very similar in their shape and their surface features match closely, but not exactly.

Each of the models had its original label, and these are shown in Figures 4 and 5. Model BM.1985,MI9558 was recorded in the Museum of Practical Geology’s Mineral Inventory register on 15 May 1911, as Number 9558, and described as “Model of a large mass of Carbonado which weighed 3073 cts or 630 gms; the original came from Bahia, Brazil”. This register shows that the source of this model was a donation from “J.R. Gregory”, and the weights in carats and grams are those reported by Gregory himself in an article published in 1895 (Gregory, 1895). These are the same weights given by Moissan (1895) who converted grams to carats using the current international standard of 1 carat = 205mg.

The British Museum (Natural History) model, Register Number BM.1911,119, is described in the historic handwritten General Register of Additions to the Collection of Minerals (vol XX part i), and on the larger of the two accompanying labels (see Figure 5), as “Model in gun-metal of the largest known piece of Carbonado weighing 3078 carats (=632 grams) found in 1895 in Bahia, Brazil”, purchased from “Messrs. J.R. Gregory & Co.” in 1911. The separate Price Catalogue (Department of Minerals, Vol. 2) indicates the purchase price of £1-1-0 (one guinea) was approved on 25 February 1911. Noted in the register, and repeated on the back of the larger label, is a reference to Gulland (1902), who reported the weight of the carbonado to be 3078 carats, and this may be the source of the weight inscribed in the Register. It was converted to grams using the then Board of Trade standard of 1 English carat = 205.304mg, giving a weight of 632g. The conversion was likely done by Leonard J. Spencer of the Mineral Department whose handwriting is in the register and is a close match to that on the small label.

The data from these registers indicate that both models were obtained from J.R. Gregory & Co. only months apart in 1911 (see Note 2). Why these models were acquired sixteen years after the find of The Sergio in 1895, remains a mystery. It seems probable that the models were made closer to the time The Sergio was found and interest in it was high.

Readers will have noticed the variation in weights given for the original Sergio in the labelling of the NHM models. Further variations are found in other references reporting on The Sergio. Some of the confusion is caused by varying conversion standards between carats and grams, which did not become internationally consistent until 1914 when the metric carat (1 carat = 200mg) was accepted by all countries including Great Britain (see Note 3). Other variations in weight occurred due to the porosity of carbonado and, being wet when first found, it gradually dried out, losing as much as 19g in the first two months (Moissan, 1895).

Description of the Models

Both models are a very dark greyish-black and an initial examination indicated they were made of a copper-coloured metal then coated with a matte black paint. Inspection of Figures 4 and 5 reveals that both models have left dark marks on their labels (also visible on the base of their boxes), suggesting that over the last 113 years some of the dark coating has rubbed off. Both models, and particularly BM.1985,MI9558 (see Figure 1b), show patches where the metal beneath has been revealed by the frictional contact.

Model BM.1911,119 weighs 680g and model BM.1985,MI9558 weighs 690g, both reasonably similar in weight to the original Sergio, which weighed about 630g. Given the size (over 80mm x 80mm x 60mm), it was clear the models were too light to be solid metal. Each model has a small hole in almost the same position on one edge, as shown for BM.1911,119 depicted in Figure 6a, which connects with a hollow centre. Figure 6b shows the matching hole in model BM.1985,MI9558, and also shows imperfections in the moulding process, with some dissimilarities in shape.

We concluded that the models were hollow casts in copper, or a metal alloy based on copper, then painted.

How Were the Models Made?

Two common methods of making hollow metal models are cire perdue, or lost wax casting, and centrifuge or slush casting. Full descriptions of various moulding methods are available in references, such as Untracht (1982) Jewelry Concepts and Technology, and YouTube videos. Briefly, in lost wax casting, a mould is made using a substance such as moulding clay to surround the original object, then it is carefully cut to remove the original, and reassembled to create a “negative” of the original. A hole is made, into which molten wax is poured to coat the inside surface to the desired thickness. The mould is broken to reveal the “positive” wax model which is hollow. This wax model is coated with heat resistant plaster, and the hollow interior is then filled with an investment material, often plaster of Paris. The whole assembly is inverted, and heated so the wax either runs out or is burned away. The assembly is then filled with molten metal which hardens to create the hollow model. The external mould is broken away, and the metal model cooled in water to allow the investment material to disintegrate and escape through the hole, which is usually then repaired. A powerful advantage of cire perdue casting is the accuracy and generally very detailed outer surface of the model. Disadvantages are that the mould is expendable, each mould can be used only once, and it is a time-consuming process.

In centrifuge or slush casting, a two (or more) part metal mould is made from the original, and keyed so it can be fitted together. Molten metal is then fed in and “slushed”, usually by a vibrating machine or centrifuge, to fully coat the interior surface and to stop gravity causing the metal to pool at the base. Alternatively, the mould could be filled with molten metal, allowed to cool slightly so the outside solidifies, then the remaining molten metal in the interior can be poured out. Slush casting was invented in 1893 for the casting of lead toy soldiers, and excess molten metal could be poured away when the interior of the mould was sufficiently coated.

For the casting of metal models, the mould is often made of grey cast iron, which has a high thermal resistance and is able to be reused. The interior mould surfaces are covered with a material (sometimes called “parting material”) that will aid the separation of the mould from the model when the mould is disassembled. Advantages of this method of casting are that the moulds are reusable, no interior investment is required or needs to be removed, and many models can be made quickly. Disadvantages are that the resulting surface cannot be as intricately detailed because the mould pieces have to be separated from the model without breaking it or the mould. Mould join marks may also need to be repaired. Readers may recall consuming hollow chocolate “Easter eggs” which are prepared using this casting method, and the join marks between the two halves are usually visible.

Which method was used to make The Sergio models? Given that both models came from J.R. Gregory & Co., and that this company sold models of diamonds (see Note 2), it is likely that the models were produced in a commercial quantity. Further, the surface of the original Sergio was not smooth, it had some hollows and raised areas, but overall, was sufficiently rounded to be easily moulded. Inspection of the models suggests that a four-piece mould was used, judging from creases or ridges that appear to be join marks. For The Sergio models to be realistic, the thickness of the metal shell needed to result in a similar weight to the original so, possibly, the appropriate weight of molten metal was poured into the mould, then slushed about to ensure that the inside of the mould was fully coated by the metal. The small hole shown in Figures 6a and 6b may represent the opening into which the molten metal was poured and the hole closed partially as the metal solidified. The hole looks too small to allow investment to escape easily, supporting the idea of models made by slush casting rather than the lost wax method.

Figure 7 shows an enlargement of the area surrounding the hole in Model BM.1985,MI9558, and inspection reveals some parallel file marks in the copper-coloured metal on both sides of the hole. indicating a “touch-up” of a join between pieces of the mould. Similar filing marks are visible on the other model. The image in Figure 7 also gives some information about the paint process. Above left of the hole is an irregular patch where the paint has flaked off, showing the bare metal. Various “imperfections” in the models, seen as slight variations of the surface features between the two, suggest differential care in the model making or perhaps even that more than one mould was used. If successive moulds were used, they were probably made by copying earlier moulds (as the original would not be available), thus later copies would be less “perfect” than earlier copies, for example, Model BM.1985,MI9558 is a little less detailed than Model BM.1911,119. How the original mould was made, and by whom, we can only guess.

The Composition of the Models

The Metal

The Register entry for Model BM.1911,119 stated that it was made of gunmetal. Today, we tend to think of gunmetal in terms of a dark bluish grey colour, but actual gunmetal is a type of bronze, an alloy of mainly copper (Cu), tin (Sn) and zinc (Zn). The standard gunmetal is an alloy usually containing copper 88%, tin 10% and zinc 2%. The melting point is around 1000˚C (cast iron is around 1200˚C) which is low enough for easy casting. As its name suggests, gunmetal was originally used in ordnance, but it has been replaced in that use by steel. Because gunmetal withstands corrosion from air and water, it is now used to make valves, pump parts, steam fittings, and bathroom tapware.

To determine the composition of the metal used in the two models, micro X-ray fluorescence (μXRF) analysis was performed using a Bruker Artax 800 spectrometer equipped with a rhodium X-ray tube. The experimental conditions were set at 50kV, 200µA, 100s live time and working distance of 12mm. The instrument was operated in air and provided simultaneous multi-element analysis between aluminium (atomic number, Z=13) and uranium (Z=92). The area examined in each experiment was approximately 100μm or less in diameter. We analysed three fully abraded spots on specimen BM.1985,MI9558 and one spot on BM.1911,119 in the areas where the dark paint coating was missing, allowing access to the metal alloy underneath. One spot on each model that was partially abraded and looked a silver colour, and another non-abraded spot on each model, where the dark paint was intact, were also analysed. Examples of these kinds of spots can be seen in Figures 8 and 9.

The fully abraded spots on the two specimens gave comparable results. The three spots on specimen BM.1985,MI9558 indicated that copper and zinc were the predominant metals in the bulk alloy; small amounts of iron (Fe) were present, and traces of titanium (Ti), chromium (Cr), lead (Pb) and tin were also seen. The relative ratios for BM.1985,MI9558 were determined to vary between 92(Cu)/8(Zn) and 91(Cu)/9(Zn). The result for the spot analysed in BM.1911,119 was similar, with a copper to zinc ratio of 93(Cu)/7(Zn).

Figure 8 shows the combined XRF spectra resulting from the overlay of the three fully abraded spots analysed on model BM.1985,MI9558, with photomicrographs of the spots 1, 2 and 3 and the fully abraded spot 6 analysed for model BM.1911,119.

The dominance of copper and zinc with traces of iron and lead are clearly seen. Traces of other metals are not labelled.

The partially abraded and the non-abraded areas give an indication of the heavier components of the surface paint. The spectral graph in Figure 9 shows the overlaid results for the fully black coated, non-abraded spot on each model, BM.1985,MI9558 (spot 5) and BM.1911,119 (spot 7), focusing on the less abundant elements rather than copper and zinc from the bulk which are still the predominant elements in the spectra. The findings suggest that, in addition to copper and zinc, small amounts of iron, potassium (K), calcium (Ca), titanium, chromium, lead and tin were present. Of these, potassium, calcium and titanium appear to be more prevalent in the black coating, probably as fillers in the paint.

Finally, to determine if the metal in the models could be described as gunmetal, three standards of gunmetal and brass of known composition, from the Victoria and Albert Museum (V&A) reference collection, were selected for analysis to allow a semi-quantitative comparison by comparing the percentage ratios of the elements in the known brass standards with the model’s bulk alloy. The brass standards were similar to the models, but the gunmetal reference was not used because it contained too much lead and tin to be comparable with the models. Therefore, despite BM.1911,119 being described as gunmetal, the two models are in fact an approximate brass composition, as true gunmetal contains significantly more tin and usually more lead.

The silvery reflective appearance of the partially abraded areas had led the authors to suspect a silvery layer between the black paint and the copper-hued metal, potentially a thin layer of tin via a process colloquially known as tinning in order to imitate the lustre of the carbonado. This was further suggested by a small piece of silver-coloured metal found loose within BM.1911,119 which was extracted from the small hole using tweezers. Analysis by μXRF indicated the fragment was predominantly tin. However, as the analysis of the non-abraded and partially abraded areas did not detect any elevated tin to suggest this thin layer, the theory was discounted.

The Paint

We examined the nature of the paint on the models using Attenuated Total Reflection- Fourier Transform Infrared (ATR-FTIR) spectrum analysis. To minimise possible damage to the models, very small samples of the paint material were removed from the specimens by scraping. The samples were loaded onto the diamond crystal of the built-in ATR-FTIR unit of an iS50 benchtop FTIR spectrometer by Nicolet (ThermoFisher) with a deuterated-triglycine sulfate (DTGS) detector. This is a very sensitive room-temperature detector for mid-infrared range measurements. Spectra were recorded from 400-4000 cm⁻¹ at 4 cm⁻¹ spectral resolution and 64 scans. Automatic background suppression and baseline correction were applied during analysis.

The resulting FTIR spectra of the two specimens were sufficiently consistent to conclude that the paint substance is generally the same on each model. The spectral graphs are given in Figure 10, and show prominent peaks at 2917, 2849, ~1700, 1461, and ~990 cm⁻¹. The peaks at 2917, 2849, 1700 and 1461 cm⁻¹ are consistent with the use of an oil or wax binder in the paint (Garrappa et al., 2022, Beltran et al., 2015). The peak at ~990 cm⁻¹ could be consistent with degraded charcoal or vine black in linseed oil (Vahur et al., 2016) but equally could be degraded bitumen, another common black pigment of the 19th Century.

The results of the (μXRF) analysis on the partially and non-abraded spots on the two models are complementary to these results. The FTIR spectral analysis was more sensitive to the binder used in the paints than the cations (potassium and calcium, etc) which were likely to be present in the fillers used in the period.

In summary then, we found The Sergio models to be made of a kind of brass, comprising approximately 91% copper, 8% zinc, minor iron and traces of titanium, chromium, lead and tin. The models have been coated with a matte black paint, in which the black pigment is a carbon black, with linseed oil as a likely binder.

Using the Models to Build a Picture of The Sergio

Our article on black diamonds (Hansen and Rennie, 2022) included two images of The Sergio, sourced from the reproduction of an engraving in an article by Moissan (1895), and a photograph taken from Furniss (1906). At the time, we surmised that one image showed the reverse side of the other, but because we found no other images, we could not confirm this assumption. The NHM models allow us to do just that. Figure 11a shows the engraved image of The Sergio published in Moissan (1895) together with the BM.1911,119 model in a similar orientation in Figure 11b. There is no doubt they refer to the same specimen. The match means we can conclude that Moissan’s engraving and Furniss’ photograph do indeed refer to the same specimen.

How well do the models represent the original Sergio? We found only two contemporary descriptions of the appearance of The Sergio. The earliest and most detailed was by Henri Moissan, a French scientist with an interest in creating synthetic diamonds from carbon, so his interest in The Sergio is understandable (see Evans, 2022, for a description of these efforts). In September 1895, about two months after The Sergio was discovered in Brazil, Moissan made a presentation about it to the Academy of Sciences in Paris and this was reported in La Nature (Moissan, 1895). At that time, carbonados were frequently referred to as “carbons”, and Moissan’s description reflects the outcomes of some of his experiments.

This new carbon, rounded in shape, is very dark black, and its surface varies from granular to smooth. The granular part, examined with a magnifying glass or under a microscope with a low magnification, has the appearance of a material which has released gases being still in the pasty [or viscous] state. It looks a lot like the surface of the microscopic carbon grains we got in our silver and iron pellets compressed by sudden cooling in water. The colour is also the same. This carbon is porous; since it was taken out of the ground, that is to say two months ago, it has lost about 19 grams in weight; at the time it was found, it weighed 3167 carats. (Moissan 1895, p.304, authors’ translation from French).

The second contemporary report was by James R. Gregory, from whose company The Sergio models were acquired. He published an article in December, 1895 in which he referred to Moissan’s presentation at the Academy of Science and wrote that the carbonado specimen “had a dark, brownish black colour, slightly polished on some of its angular surfaces, as is usual with examples of this variety of diamond” (Gregory, 1895, p.1536).

These reports indicate that The Sergio’s colour was black, perhaps with a tinge of brown. The surfaces were smooth to granular, and some appeared to be “polished”, so we can assume some reflection of light.

The NHM models are black, with a hint of brown, and some reflective surfaces (although the possibility of wear through handling cannot be discounted). The models’ surfaces vary from smooth to grainy, suggesting that considerable care was taken in painting the models. Certainly their shape, size and weight closely match those of the original Sergio. We decided that the models were a good approximation of the original Sergio, and as The Sergio was lost forever, we thought that this was the nearest version of it we could hope to see.

We wrote our brief description of The Sergio and its history, based on the information we had uncovered that seemed to be credible (Hansen and Rennie, 2022). It was found on July 15, 1895 by a miner named Sergio Borges de Carvalhoo, from whom it was first bought and not long after, at an unknown date, it was broken up and sold. At that time we had found no other confirmable information.

Finding the models gave us some new leads to follow and discover more about what happened to The Sergio after its discovery. In renewing our research into the history of this specimen, we were assisted by the references recorded in the General Register of Additions to the Collection of Minerals (vol XX part i). Fortunately, the archives of the NHM library are remarkably vast, and the search facilities available on the internet today reveal a great number of reproductions of historic printed material.

We set off on a new journey of historical research that took us from Brazil to Paris to London, Then, after a surprising find in an obscure literature source, we went back to Paris in the present time in search of an original model of The Sergio, commissioned by Moissan himself. We wondered if it was located in the mineral collections of the Muséum national d’histoire naturelle, Paris. The conclusions to this second part to our journey to uncover the life and times of The Sergio will appear in the next issue of this journal.

Notes

Note 1. The Museum of Practical Geology was founded as the Museum of Economic Geology, opening in 1837 in Charing Cross. In May 1851, it reopened as the Museum of Practical Geology, initially sharing its building with the Geological Survey, the Royal School of Mines and the Mining Records Office. In 1935 the collections of the Museum of Practical Geology moved to the newly constructed Geological Museum building, next to the then British Museum (Natural History) in Exhibition Road, South Kensington. All of the mineral and gemmological collections were amalgamated into the Natural History Museum, London in 1985.

The Natural History Museum has its roots in the British Museum, founded in 1753. As the British Museum’s collections grew too large for the premises in Bloomsbury, the natural history items were separated and moved into the purpose-built building on Cromwell Road, South Kensington, opening in 1881 as the British Museum (Natural History). This museum separated from the British Museum in 1963 and was officially renamed in 1992 as the Natural History Museum.

Note 2: James Reynolds Gregory (1832-1899) was a well-known geologist, mineralogist and dealer in London from 1858 until his death in December 1899. Cooper (2006) reported that James was a member of the Geological Society, the Mineralogical Society, and of the Society of Arts in England, as well as the Mineralogical Society of France. He changed the name of his company from James R. Gregory to J.R. Gregory & Co. in 1896, recognising that his eldest son Albert G. F. Gregory, who had been assisting him for about 15 years, would continue the business. During his time as a dealer, James also offered classes in geology and mineralogy, and sold various kits and equipment as well as geological specimens of every kind. Cooper (2006, p.146) reproduced a full-page advertisement from James R. Gregory published in Nature, in 1877, that advertised “models of crystals and exact facsimiles of 24 of the large diamonds”. It is therefore not surprising that models of the large carbonado could be obtained from J.R. Gregory & Co. However, when the two carbonado models pictured here were obtained in 1911 from the company, it was Albert, rather than James, who was in charge.

Note 3. The carat, as a unit of weight for very small objects, has a centuries-old history. It likely began as the weight of a seed from the carob tree (Ceratonia siliqua), whose seeds were considered to be remarkably consistent in weight. Recent research has demonstrated this is not the case, with seeds varying by as much as ±20mg around the metric carat weight of 200mg (Turnbull et al., 2006). Not surprisingly, the weight of the carat varied considerably across countries. Webster noted that at one time the value varied from 188.5mg to 213.5mg (Webster, 1975, p.437). One attempt to standardise the carat weight occurred in 1871 when jewellers in France agreed to set 205mg as the international standard for one carat, but other countries persisted in using their own standards. For example, in 1888 the Standards Department of the Board of Trade in England confirmed the English carat to be 205.304mg. Zhengzhang (1991) also investigated the variability of carob seed weights and listed twenty-three different measures prior to 1907, ranging from 187.00mg in Cyprus to 215.99mg in Livorno.

The discovery of large diamonds, such as the Cullinan in 1905, increased the necessity to standardise the carat – Spencer (1910) lists eight different carat values of the Cullinan’s weight published between 1905 and 1910! Spencer was Assistant in the Mineral Department at the British Museum (Natural History) (and later the Keeper), and his conscientious efforts to determine its exact weight before cutting highlights the complications inherent in the diversity of carat values. In 1907, the French Comité International des Poids et Mesures proposed a metric carat of 200mg that was accepted by the Quatrième Conference Generale des Poids et Mesures in Paris. The United States adopted the metric carat definition on 1 July, 1913, and it was legalised in the United Kingdom on 1 April 1914 (Kunz, 1917).

References

Beltran, V., Salvadó, N., Butí, S. and Cinque, G., 2015. Micro infrared spectroscopy discrimination capability of compounds in complex matrices of thin layers in real sample coatings from artworks. Microchemical Journal, 18, pp.115-123. DOI.org/10.1016/j.microc.2014.09.001.

Brazil. Commissão, Louisiana Purchase Exposição, 1904, and F. M. De Souza Aguiar. Brazil at the Louisiana purchase exposition. St. Louis, Saml. F. Myerson Ptg. Co. [online] Available at <https://lccn.loc.gov/18006587> [Accessed 16 January 2022].

Cooper M., 2006. Robbing the sparry garniture. Mineralogical Record, Inc. Tucson, AZ.

Evans, J. 2022. The early history of synthetic diamond. The Australian Gemmologist, 28(2), pp.90-95.

Furniss, H., 1906. Diamonds and carbons in Brazil. Popular Science Monthly, 69(9), pp.272-280. [online] Available at <https://en.wikisource.org/wiki/Popular_Science_Monthly/Volume_69/September_1906/Diamonds_and_Carbons_in_Brazil> [Accessed 14 January 2022].

Garrappa, S., Frøysaker, T., Streeton, N.L.W., Hradil, D., Platania, E., Beltinger, K. and Caruso, F., 2022. Micro-spectroscopic study of late 19th-early 20th century tube paints. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 279(121414). DOI.org/10.1016/j.saa.2022.121414.

Gregory, J., 1895. A remarkable diamond. The Mining Journal, Railway and Commercial Gazette, 65(3147), p.1536.

Gulland, J., 1902. Brazilian carbons. Journal of the Society for Arts, 51(26098), p.22. [online] Available at <https://www.jstor.org/stable/41335675> [Accessed 21 September 2022].

Hansen, R. and Rennie, L., 2022. Black diamonds: A reflective review. The Australian Gemmologist, 28 (1), pp.26-36.

Kunz, G., 1904. Precious stones. In D.T. Day ed., Mineral resources of the United States, Calendar Year 1902, pp.813-865. Washington: Government Printing Office.

Kunz, G., 1917. The international language of weights and measures. The Scientific Monthly, 4(3), pp.215-219. [online] Available at
< https://www.jstor.org/stable/22412> [Accessed 10 August 2023].

Moissan, H., 1895. Carbon noir du Brésil. La Nature Revue des Sciences, 32(1166), p.304.

Spencer, L., 1910. Notes on the weight of the ‘Cullinan’ diamond, and on the value of the carat-weight. Mineralogical Magazine, XV(71), pp.318-326.

Turnbull, L., Santamaria, L., Toni Martorell, T., Joan Rallo, J. and Hector, A., 2006. Seed size variability: from carob to carats. Biology Letters, 2, pp.397-400. DOI:10.1098/rsbl.2006.0476.

Untracht, O., 1982. Jewelry: concepts and technology. London: Robert Hale.

Vahur, S., Teearu, A., Peets, P., Joosu, L. and Leito, I., 2016. ATR-FT-IR spectral collection of conservation materials in the extended region of 4000-80 cm^–1, Analytical and Bioanalytical Chemistry, 408, pp.3373–3379. DOI.10.1007/s00216-016-9411-5.

Webster, R., 1975. Gems, 3rd ed. London: Butterworths.

Yawger, I., 1907. Carbon: a remarkable mineral, from The Mining World [online] Available at <https://babel.hathitrust.org/cgi/pt?id=njp.32101051619235&view=1up&seq=1 [Accessed 19 December 2022].

Zhengzhang, T., 1991. On the origin of the carat as the unit of weight for gemstones. Chinese Journal of Geochemistry, 10(3), pp.288-293.