Metal-detectorists Reg Mead and Richard Miles had been investigating the same field on the Channel Island of Jersey for three decades, recording occasional stray coin finds throughout this period. In May 2012, though, they discovered a rather more substantial concentration and, realising the significance of what they had found, immediately refilled their hole and contacted Jersey Heritage, the professional museum body on the island. A team of local archaeologists, with me as conservator, was assembled to investigate the discovery, but we little suspected how important a find this would prove to be. When we began our excavation four weeks later, we only initially opened a 2m by 2m hole because we were expecting to uncover a small hoard of perhaps a few hundred coins, possibly in a pot. In fact, it soon became apparent that the hoard was much larger – indeed, it wasn’t until the end of the second day that we managed to expose all of its edges. Thousands of coins, fused together by a layer of green corrosion, formed a mass measuring approximately 140cm by 70cm, and 15cm thick. Even at this early stage, it was possible to identify many of its contents as silver/copper alloy staters and quarter staters made by the Coriosolitae tribe of Celts from the neighbouring French coast, and dating to around the middle of the 1st century BC.

Given the hoard’s clear importance, it was decided to excavate it intact in a single block. This was no small task: in order to keep the coins damp during their recovery, a thin (10-30mm) layer of earth was left on the hoard’s top and side surfaces, and we then dug down below the level of its base all around it, leaving the mass on a layer of earth about 200mm tall. This was to act as both a supporting base and as a reservoir of water in order to buffer the hoard itself. As an extra precaution, hoard and base alike were wrapped in cling film when not being worked on. The next stage involved particularly delicate work: to ensure the hoard’s safe extraction from the trench, we needed to reduce the hoard’s contact with the ground as far as possible, so that it wouldn’t be stuck hard to the earth underneath it when the time came for the lift. To that end, the earth layer beneath the coin block was removed as far as it was believed safe to do so, leaving it intact for the full width of the hoard where it touched the coins, but tapering down like an inverted cone. Hand tools were then used to dig four evenly spaced parallel tunnels, each 120mm in diameter, through the earth that remained under the hoard.

It was then time to bring in rather heavier equipment: the hoard and its surrounding earth weighed around a ton, so it would only be leaving its 2,000-year-old hiding place with the help of a crane. A metal scaffold structure was constructed to fit closely around the hoard, with four nylon belts fitted through the tunnels underneath it and around the scaffold so that it could all be lifted out by crane. The belts would support the hoard’s weight as fully and evenly as possible, while the crane’s chains and hooks touched only the metal structure. Layers of foam and plywood shielded the hoard’s top edge from being compressed by the belts, and once the block had been raised about 30cm, another foam layer on a specially constructed wooden pallet was placed beneath. Finally, this protective sandwich was lifted by an extending crane on to a flatbed truck, and the hoard was carefully transported to our laboratory for further investigation.

Initial investigations

In the first two years after the discovery of the hoard (now known as Le Câtillon II; the first Le Câtillon hoard, found in 1957, comprised c.2,500 Celtic coins), Jersey Heritage was granted permission both to remove overlying earth to reveal its surface fully, and to perform non-invasive research to reveal more about its make-up, but we could not at this stage take apart the mass of coins to examine them individually. We discussed the possibility of X-raying the entire block with colleagues at Southampton University, but the mass of coins was too large for the available equipment. Limited X-ray fluorescence (XRF) analysis of exposed surfaces was undertaken by Karl Harrison and Andrew Shortland of Cranfield University, however, and we were also able to carry out some microscopic studies of organic materials in the block’s surface, revealing various plant and animal remains.

What quickly became clear, though, was that the contents of the hoard were more varied than at first thought: partially exposed gold torques, silver bracelets, pieces of flat gold sheet, glass beads, and many other artefacts were identified scattered amid the coins on its surface. As a whole, the assemblage was such an impressive object that we initially considered preserving the whole block intact, but in light of its eclectic make-up it was decided that we should instead disassemble the hoard on an object-by-object level in order to reveal all of its contents.

Once funding had been obtained, this work began in 2014. It was a painstaking process, conserving and recording each piece as it was removed from the block, and, while the conservation work followed conventional best practice, it was complemented by an innovative method of recording the finds using a cutting-edge piece of kit called a Faro Edge six-axis metrology arm, which had a contact point probe and a scanning laser head. The laser was used to record scans of the hoard’s surfaces and its contents at regular stages of the work, while the physical contact probe head was used daily to record the position of every single coin and all the other finds on the hoard’s surface to fractions of a centimetre in three dimensions. This meant that, even after the hoard had been taken apart, future researchers would have a virtual model of the mass as found, in which the coin positions would be linked to their database records of tribe, chronology, and so on.

This was a key theme: we were keen that our work should allow the maximum amount of research, both during the disassembly phase and for many years to come, so we took advice too on the recognition, extraction, and preservation of samples of interesting soil, organics, and corrosion that we encountered during the process. Nor did we deconstruct the entire block: a solid, full height 15cm by 15cm by 15cm section of the hoard was left intact with all its earth and organics remaining in place between the coins, while 1,500 more coins – still only a small portion of the whole – were left untreated as a research archive. There will be plenty for future researchers to study, giving them the benefit of future advances in research techniques.

In that spirit, we began our work. The team comprised three professionals – two conservation assistants, Viki Le Quelenec and Georgia Kelly, both of whom were qualified archaeologists, supervised by me as conservator – and we were supported by a diverse range of professional advisors and 20 part-time volunteers who did much of the hands-on work. Olga Finch, the archaeology curator for Jersey Heritage, acted as the archaeology/research advisor, and the two finders, Reg Mead and Richard Miles, acted as principal coin identifiers, their work overseen by Philip de Jersey, Guernsey Museum Service archaeologist. So, how did the work progress?

Cutting-edge technology

The hoard block was disassembled one coin at a time. This process involved a team of two people: one to physically remove the coin and operate the metrology arm that measured the coin position, while the other operated the computer linked to the arm and assigned a sequential number to each new coin, before bagging it. Plotting each coin only took a few seconds, touching the point probe three times across its centre axis, and three more times around its edge, to position it in three dimensions using Geomagic software – using this method, little by little we were able to build up a three-dimensional map of the coins. We also used laser scanning to record the position of newly uncovered torques and other items, so that they could be placed precisely in the final three-dimensional virtual hoard model.

Helpfully, the act of physically separating the coins proved easier than had been envisaged. They were connected to each other only by the fusion of their copper corrosion products, and this bond fortunately proved weak – it was always possible to remove a coin one at a time using a metal hand tool, usually a specially blunted knife, without causing any damage. As for the other artefacts, sufficiently robust jewellery and other objects were cleared of all surrounding coins and then the number around their base was also reduced. At some point in this process, they usually came loose, and if not they would be gently manipulated by hand until they came free.

As the hoard was potentially full of important organic material, we had to keep it wet at all times. This was done in two ways. At night, or during longer periods when the hoard was not being worked on, it was kept tightly wrapped in layers of cling film. The hoard spent the majority of 2012-2014 swaddled in this way, kept in a cold store at 4ºC, and during the following period, when it was being excavated in a publicly viewable open lab, it was kept wrapped at night but had to be actively moistened when it was on display. While working on the hoard, we made use of hand sprayers containing de-ionised water, but when the mass was left alone for any length of time we set up an ultrasonic humidifier to produce a stream of cold, de-ionised water-saturated air that exited from a series of holes in a special device positioned along one edge of the hoard so that the vapour slowly rolled over its surface.

The size of the hoard – and the task of conserving all of its contents – was unprecedented, and it was decided early in the project that the coins would be cleaned only enough to allow identification, rather than being polished to display standard, due to the sheer number that we would have to process. After consulting the British Museum’s conservation team, who had been working on a smaller hoard of similar coins, we did this by using 15% formic acid to remove the copper corrosion products from the silver/copper alloy coins, followed by a sequence of six baths of water to remove any traces of the acid. We soon worked out an efficient system that allowed us to treat around 120 coins per day. Given that the coins numbered in the tens of thousands, you can imagine the scale of the project before us.

The recording and tracking of each coin through this process was particularly important. When removed from the hoard, the coin’s number was written on its individual polythene bag, and the coins underwent treatment in a modified ice-cube tray. Each tray had a different identification letter, and each of its compartments had a number – thus we might know, for example, that coin CATII/H/23671 was in tray B, compartment 7 – and these allocations were carefully recorded on paper records and Excel spreadsheets.

After they were washed and dried, each coin was then transferred to a new polythene bag, marked with its individual number and in-hoard position. This record-keeping has so far provided us with a grand total of more than 69,350 coins (the exact number is still uncertain because we can only estimate the number in the block that was kept intact for later research), which makes Le Câtillon II easily the biggest Celtic hoard ever found, as well as the largest coin hoard yet found in the British Isles.

As for the other artefacts, inorganic objects other than coins were divided broadly into the following types: gold jewellery, including a number of torques; gold sheet and scrap; silver jewellery and wire; silver alloy ingots; and glass beads. The treatment of gold and silver jewellery had two phases during the project, initially with chemicals and later typically without, but in either case work on each artefact began with the removal of loose earth and corrosion using just water and soft brushes. The gold artefacts often had copper corrosion products on their surfaces, which came from the surrounding coins in the hoard; this was manually removed under a stereomicroscope using Berberis thorns held in a small hand tool. This was a tip from conservators who had worked on the Anglo-Saxon Staffordshire Hoard (see CA 290), as the thorns were robust enough to remove most of the corrosion, but soft enough not to scratch the gold surface. This removed the vast bulk of the material, but a thin green corrosion glaze sometimes proved impossible to remove manually. Initially, we very carefully used dilute formic acid solution on cotton wool swabs to remove it, and this worked very well. Later on, however, it was decided to stop at the manual cleaning point and not use the acid, in case it might affect future metallurgical research. Glass beads were simpler to clean: we usually just washed them in de-ionised water, and if there was any copper corrosion product adhering to their surface, this could be gently removed with a scalpel.

In addition, there was a partially surviving leather purse or pouch that was still full of coins. This was excavated as a block from the surrounding hoard and removed on a corrugated plastic sheet, where all possible coins were then removed from its exterior. The block was next treated by freeze-drying with PEG (polyethylene glycol – the substance used to preserve the timbers of the wreck of the Mary Rose). Untreated, the leather would have shrunk, twisted, and hardened on drying, but the PEG permeates and bulks out the leather, allowing it to dry safely (untreated fragments of the purse were kept for analysis, too). Once dried, the leather surface was brushed clean of remaining earth and consolidated with an acrylic emulsion. Similar treatment was given to a very fragmentary textile container full of gold and silver jewellery and metal fragments, which was excavated in a block and freeze-dried, though without the use of PEG.

Scientific analysis, past and future

With the hoard largely disassembled and its contents conserved, we have now been able to begin analysis of its components. All of the gold torques, both partial and complete, as well as other items of jewellery, have been X-rayed – initially in the Jersey General Hospital, and then at higher resolution in the conservation department of Birmingham Museums Trust (which is also home to a bench-mounted X-ray fluorescence unit that we used on the majority of the jewellery).

We also took more than 900 samples from the organic materials that had been trapped between the hoard objects; these were placed wet in polythene bags and stored in air-tight containers in a refrigerator for future research. The majority were plant fibres and stems, but we discovered a number of insects, too. The first two millipedes that we identified were informally dubbed Arthur Pod and Millie, but as more were discovered we reverted to using numbers to record them – all the excavated arthropods have now been identified to family, if not species, level, and more work on them is ongoing.

There is still much to do, however. When we began the three-year process of disassembling the hoard, we assumed that that would be the entire project. It became obvious very quickly that the hoard was more important than we had imagined, and that it had unique potential for further research into this period of Jersey’s history. In 2016, therefore, we held a conference and made plans for a second phase of investigation, into both the hoard itself and its wider context.

As has already been mentioned, some work has now begun on this second phase, including X-ray and XRF analysis, as well as examining the organic specimens, while further metallurgical research has recently been carried out at Orleans University as a part of the Europe-wide ‘Celtic Gold’ research programme. Our current priority, in order to enable more research, is to complete the linking of in-hoard object positions to the database containing all the information about each individual artefact. An initial lower-resolution linkage of the two databases has already been made and, intriguingly, it appears to show that the Câtillon II hoard as we know it may have been formed by the pouring of two previously separate hoards into the same pit. When Richard Miles, one of the finders of the hoard, investigated the linked databases, he found that the hoard was composed of two distinct coin populations, one lying on top of the other. One of these layers appears to contain older, higher-quality coinage and all the jewellery, while the other contains only the later, poorer-quality coinage.

We will probably have to wait until we can properly investigate the hoard site itself and the surrounding area to gain a clearer idea of just why the hoard was there. Although we found some building material in the initial field excavation, we do not know if the site was occupied, perhaps by a village or even a hill fort. Interestingly, Le Câtillon, the name of the location, means ‘castle’, though there has never been one there in recorded times. We know other hoards of the same period have been found close by, but we don’t know if they were intended to be recovered or were perhaps offerings. Finally, we don’t know why Jersey appears to be so uniquely tied to these apparently French hoards, as it is worth noting that most French Iron Age coinage has actually been found on the island, not the French mainland. We hope that research in the coming years will answer some of these questions.