Coriosolite Expert System banner
Home General Index Index of scientific subjects
The Stylistic Analysis of Celtic Coins
By careful analysis of style in Celtic coin series we can avoid many classification errors that occlude our understanding of distribution patterns. We can also construct chronologies where die links do not exist, and provide more accurate sets for testing.

While improving the data for numismatic procedures is laudable, and first encouraged me to embark upon these methods, I soon discovered that this data also began to provide answers to many problems of Celtic art and symbology.

Regional styles of Celtic art have been identified, but the archaeological record rarely provides us with enough objects that can be attributed to a single individual and be arranged chronologically. Coins are an exception to this. Not in every case, though, as some series are so conservative that few changes in design occur, and these changes might be arbitrary. But where there is a rich vocabulary of design elements, such as in Armorica, individual thought processes and general tenets of Celtic art can be detected.

As a subject, Celtic symbology is especially prone to fanciful interpretations, especially when we are dealing with pre-Roman imagery. By understanding the structure of the art, and how the Celtic artist uses symbols, we can amass a large amount of evidence for specific interpretations that would be impossible through other means.

1 - Selecting the Series

Before attempting any stylistic analysis, we must decide if such an analysis is likely to be useful. It is not enough to state that as a particular series of coins has been well studied, there is little point in going over the same ground. Many errors of classification persist unnoticed because the results appear to be right. There is often a high degree of tolerance allowed for certain procedures, and when we deal in averages, results due to randomness, or sampling errors, can be easily attributed to other causes. Thus some hypotheses might be better stated as justified, rather than proven.

The sole criterion in selecting a series of coins for stylistic analysis is that there is enough data to work with. We are looking for variety in design. This variety should exhibit an effort on the part of the die engraver to develop the design along artistic and symbolic lines. A crude issue, consisting of only a few dashes and dots would not be suitable, regardless of the varieties in the number of such elements throughout the series. We could not state that the die engravers wished to increase or decrease the number of elements. We might find that these numbers varied up and down. Similarly, it is usually a futile exercise to count the number of hairs on a head. Rybot did this for the Coriosolite series. When I had worked out the die chronology, I checked this feature and found that there were an optimum number of hairs on most of the coins to within two or three. The earlier coins generally had less, but once the engravers became practiced, they were able to get a reasonable number into the space allowed by the rest of the design. All through the series, there were exceptions, and nothing could be predicted about the chronology through this feature.

The most valuable data will be obtained from small details, as these have a lesser chance of being symbolic and thus changed for reasons other than artistic improvement. Small symbols might signify something else, and if we have only these to work with, there is little chance of success. We should look for different ways that the same detail is engraved, as the horse's ears, mouth, eyes, tail, or hooves, that might provide some variety within a series of very similar coins. The mouth of the head on the obverse might show another series of design changes. Symbols should be examined as well, but the point is not to rely upon these alone.

We should be careful in determining what constitutes a set at this preliminary stage. If we have selected too much, and have included two or three series, these will sort themselves out in the analysis, but if we have omitted certain coins, we might never discover them later. It is easy to make assumptions about the coins: for example, if we have a tribe's coinage isolated by type, but the coins have a series of names inscribed upon them, we might assume them to be of successive kings, and analyze the coins of each king separately. This might turn out to be valid, but as it is known that certain areas had two or even three joint-kings, perhaps we should look at them as clan leaders. To avoid this problem, the names on the coins should not form the most important divisions in the chronology, unless supported by the evidence of other design elements. Even with this other evidence, there is a possibility of different mints, each having its own die engraver, and thus a new chronology existing for each mint. How each issue related to the others would be a difficult but not necessarily impossible problem.

At all stages of stylistic analysis, other numismatic evidence should be referred to, and compared with the picture we are building up. An open mind should be maintained at all times about the evidence, and the subject is better served if it is allowed to reveal information itself. Hypothesis testing will only work if you have predicted the truth about something: it is pointless to fret over the lack of hypotheses, when there is much raw data to accumulate. All you really need are your powers of observation. Analytical thought is not required at this stage.

By a happy coincidence, the most thorough and accurate die reconstructions in Celtic numismatics were performed upon what probably will turn out to be the most varied series. That Major Rybot lived a short drive from where the largest hoard of these coins were unearthed, and that he had the interest, patience, and skill to draw these reconstructions is the essence of this coincidence. Other series may fall far short of the Coriosolite coinage, but even a few changing details can provide a chronology, and also answer other questions about the art and symbology of Celtic coins.

2 - Preparing the Data

The first step in the stylistic analysis of a series of coins is the reconstruction of the die designs. Most Celtic coins do not have complete impressions of the dies; the flans are often too small; wear, and poor striking also limit the amount of detail on a coin. Sometimes we find that the peripheral design is distorted, and corrosion can obliterate important details.

Photographs can be inadequate for detailed study if they are improperly taken or reproduced. One-to-one (1:1) scale photographs, of high contrast, and subsequently printed by the half-tone process, are often almost completely useless. I have had success photographing coins with a macro lens in sunlight against a mid-dark grey field with a slightly warm tint, using a 100 ASA colour film that is not designed for portraits. This brings out the colours and details of most coins.

The angle of the coin against the light should be adjusted to give the best results. A raking light is best for coins that are badly worn and of low relief, while on high relief coins, this will create dark shadows that can hide details. It is important, when photographing the coins, to keep the focal length of the camera at the same distance. Some coins may have flans of varying thicknesses, and these will be brought into crisp focus by raising or lowering the coins accordingly. A table that can be adjusted precisely is best for this, but placing thin cards or sheets of paper beneath the background works as well. It is important that the angle of the lens be at the same angle as the face of the coin. Again, a table that is adjustable thus will be useful, but cards can be placed underneath the background at whatever side needs to be raised. The prints obtained should be on paper that has no visible grain, and be at least of 2:1 scale, or more if the coins are very small.

Colour is rarely used for photographing coins as it fades over time, but colour is very useful here, more easily defining shapes caused by corrosion, and helping the die-reconstructions that are the important part of the record. Colour is also now cheaper to process, and the technician producing the prints will be better able to adjust the colour if he is given the background to use for colour comparison. Slight variations in the reproduced backgrounds do not badly affect the quality of the photographs.

When Major Rybot made his die reconstructions of the Coriosolite coinage, he did so by drawing over the photographs with India ink, and then bleaching out the photographs. Today, it is better to trace the designs with ink on Mylar. This preserves the photographs so that they can be rechecked if needed, and provides transparencies that can be superimposed to compare different coins. The photographs and the transparencies should all be cross-referenced to the coins, and labelled accordingly.

As many coins should be recorded as are necessary to reconstruct the entire design. Sometimes a coin might be unique, or a few coins all lack the same detail. It is best to leave that detail out rather than guessing its form. After a chronology is constructed, the missing detail might be apparent by the same detail on the two coins on either side of it in the chronology. If the reconstruction is almost certain, it should be drawn with a broken line, to signify its tentativeness. Rybot also used a thicker line to show high relief, as this gives the impression of both volume and shadow.

There are many that would find this work too demanding, or would be unable to produce a satisfactory drawing. While it does not take the type of skill needed to draw from life, it is, nevertheless, a task requiring some artistic and technical skill. The width of all of the lines, for example, even those drawn to represent volume, should be on the same scale. A commercial artist would be familiar with all of the steps required, and a minimum amount of instruction on what not to reproduce, would lead to professional results.

After the die reconstructions have been accomplished, they should all be photocopied, and pasted up with obverses and reverses together. Rubber cement should be used for all pasting as it will not wrinkle the paper, and can be rubbed off cleanly. Excessive rubbing might fade the impression and should be avoided. Depending on their size, the reconstructions could be reduced before being pasted up, with care that the reduction does not eliminate any details, or make them difficult to study without magnification. There is really no point in having 1:1 scale illustrations: it is merely a convention that inhibits the study. If you are unable to tell that a coin that you are holding is the same as one illustrated 2:1 scale, there is little point in attempting to analyze the style.

Two or more copies of each reconstruction should be made. One copy of each of the sheets depicting several coins with their reference number placed between obverse and reverse, should be pasted on to thin card to facilitate handling. When this is accomplished, the coins can be cut out with a razor knife, making sure to keep obverses and reverses on the same card. The other copies should be retained on paper only, as an index of design elements can be constructed from them later.

3 - Viewing the Designs

A considerable amount of time should be devoted to observing the details of each die reconstruction. If you are the person who drew them, some of this observation has already occurred, but if someone else prepared them, and they are complex designs with many different details, then be prepared to devote days, if not weeks or months, in familiarizing yourself with every detail. It will amaze you to discover how much can be missed after looking at the same design a number of separate times. As we are looking for relationships between the designs, studying each design for fifteen minutes, and then going on to the next, will not be as effective as studying each design for five minutes, and then repeating the process from start to finish three times. Even if you are not aware of the process, the mind assimilates much of the information by repetition. Look at the designs the last thing at night, and again the next morning. Try studying them at different times of the day.

Do not look at each design element for itself alone, but try to see how it relates to other parts of the design. Celtic artists were masters of composition. They knew how to lead the observer's eye from one part of the design to another. Perhaps one of the finest examples of this can be found on a sword scabbard where interconnected registers of decoration appear to be duplicated down the length of the scabbard. Upon closer observation, it can be seen that each register is different, yet the balance of each is so harmonious, it is only this, and the repetition of registers, that we observe.

To avoid confusion, some terms should be defined:

Type Refers to the subject of the design, for reasons of general identification, e.g. "boar type" "face/horse type".
Style The way that types are depicted, not the types themselves. This can be applied specifically to an individual die engraver, or generally, to a region, e.g. "Armorican style".
Motif Any subject within the design. It could be the horse, the boar beneath it, or the symbol above it.
Design-element A part of a motif that might be recombined with others to form other motifs. It is a discrete element that is not usually broken down any further. While a pellet can be a design element on its own, if it is repeated to form a beaded line such as that forming the horses mane, the mane itself can be called an element. A curl at the end of a line is called an element as other curls might exist on lines of different shapes. Although two curls exist on different motifs, the eye associates the similar shape of that part of each motif. The designation of design element or motif might vary, depending on its context.
Echoes A design element that is repeated by either an identical element, or by a second element that contains some reference to the first, e.g. "(" echoes "(", and "(" also echoes "((", or even "()".
Opposes A design element that is a mirror image of another design element, or has some similarity, as in the previous example, e.g. "(" opposes ")", and "(" opposes "))", or even ")(". The terms "echo" and "oppose" refer to elements or parts of elements that are closest to each other, e.g. "(" opposes, rather than echoes ")(" in the following: ( )( .
By being aware of how parts of the design relates to other parts, how one design element can echo, or oppose another design element, and how the eye is led around the composition, a feel for the art can be obtained, and the motives of the artist soon become apparent. This can be invaluable to the task of establishing the chronology, as it is often necessary to decide if a design is being improved, to discover the direction of the chronology.

4 - Sorting the Data

Taking the cards showing each of the die reconstructions, we proceed to group together similar elements that are in the same positions. The best elements to choose are those that change their form frequently. Once a group of identical elements have been gathered and laid out, observe the general designs of each. If they contain two or more differences in other elements within the group, then place these together. Compare what you have selected with the rest of the die reconstructions.

You might find that similar styles or devices exist here and there throughout each group, and that no pattern is emerging. This is not reason for alarm. If the most obvious elements can be sorted easily, others would probably have noticed this already and established these elements as the defining factors in various classes of the coinage. Start again selecting different elements. This is a frustrating procedure at times, but perseverance can pay off.

Symbols are often duplicated at different parts of the chronology. These can be selected by the die engraver arbitrarily. They often cannot be used as the basis of a chronology. If you found that you had coins of different characteristics appearing in all of the groups you separated, you might discover that these are symbols. Another reason for apparent failure in the early stages of sorting the designs is that some die engravers revert back to previously abandoned elements. They do this in moderation, however, and you will usually find that other elements in these coins have been evolved.

Once you have isolated a design element to a specific run of dies, and these dies resemble each other more than they do many other dies, then use these as your master set, and find changes within these designs to other elements. group the secondary elements together so that you have a group containing two or more sub-groups. Search the other cards for more examples of the secondary elements. With luck, you should find that two further groups can be constructed that will attach by the same elements to each end of the sub-groups. You will now have defined four or more design groups.

If you found that you split your first group into three sub-groups, and that only two of the design elements were represented elsewhere in the cards, then the missing element might be placed between the first and last.

The direction of the chronology that you are constructing might be difficult to detect. If no definite improvement to the design is obvious in either direction, other numismatic evidence might at least form the basis of a working hypothesis. This is usually only a problem in series that have little variation of design.

At each stage of the process check other design elements and look for similarities in the "look" of the coins. As die engravers work, they adopt certain habits that are repeated, and some of these are barely definable. They might be slight similarities in the proportions, or in the angle of a nose, they cannot be relied on, but can serve as indicators for further examination. All definitions of chronological positions should be irrefutable, there is no room for personal opinion here.

Some coins will have to be grouped together where their internal chronologies cannot be established. It is only here that opinion can enter the picture. I always mark such runs in a catalogue with the use of parentheses, as in the following example:

Coins: 1, 2, 3, (4, 5,) 6, 7, 8.

The relative order of Coins 4 and 5 is unknown, and they are placed in that order either arbitrarily, or by personal judgement. The relative positions of the parenthesis and the other numbers is certain.

5 - Constructing the Flow Charts

By plotting each design change on flow charts, chronologies, and breaks in the sequence denoting different issues, can all be seen at a glance. The patterns of such flow charts identifies, and characterizes individual die engravers.

As soon as some order has been established for a number of cards, the information is plotted on graph paper, listing the reference numbers of the die reconstructions along the top, and the elements or motifs with their changes in order down the left side. It is best to do this in pencil, so that adjustments might be made as the chronology is developed. Separate sheets can be used for obverse and reverse design elements. In this way, newly discovered elements can be plotted beneath the last entry. Once all elements are recorded, then the sheets can be pasted together or redrawn.

Drawing a new chart for each grouping, is unnecessary, but sorting out too many features without recording the details, can lead to confusion if you forget some of those details. You will find that while you are focused on one part of the design, you can easily miss details on another part. Do not be dismayed by this, you might be duplicating some of the engraver's thought processes: I have noticed that an engraver can focus on one part of the design and make changes to it, while neglecting another part, only to work on the second part after he is satisfied with his first results. This is especially noticeable where the attention is alternately switched from obverse to reverse changes. However, the best die engravers seem able to change as many as three or four parts of the design at nearly the same time.

To avoid becoming too fixated; take many breaks in the task so you might view the designs freshly. Of course, if you are working on a series that has only two or three changes in the designs, much of this advice is needless. The Coriosolite coinage took over 120 changing design elements plotted against 29 motifs, and used about 200 die reconstructions.

After plotting the first coins, the chart might look something like this:

As the positions are still tentative, there is no point in changing the numbers. Two features have been charted, with three differences in the first and two in the second. This part of the chronology might end up having more features. It is the overlaps that define the chronology. The first five coins can be defined as 1.1/2.1, the next three as 1.2/2.1, the next four as 1.2/2.2, and the last seven as 1.3/2.2. This gives four chronological steps with a reasonable expectation of accuracy.

We might not be sure which direction the chronology runs. Only very obvious improvements in the design might establish this if there is no other numismatic evidence. This part of the chart might be adjusted in the light of evidence from other features, for example, it sometimes occurs that the die engraver forgets a previous change, and reverts back to the abandoned element for one die, or he may even revert back to an abandoned element for many dies if he becomes unhappy with the prior change. In some cases, a die engraver recombines previously used elements with new ones, presumably to affect some sort of originality on the design without inventing too many new elements. All of these traits will be revealed as more features are taken into consideration.

Building up a complete chronology can be very difficult, and takes patience. Each die engraver has his own idiosyncrasies, but these can be studied as well. The individual styles and smethods of working can give us new insights into the tenets of Celtic art generally. Three lines of a finished flow chart might show the following sort of pattern:

The first line is typical of those that might precede it, but the last two lines show reversions to previous elements. These two patterns are the commonest types of reversions. The first is likely to be some sort of symbol, where the design can be changed without affecting the meaning, for example, sun symbols of varying forms. The second reversion ise of the enduring kind and signify a change of mind for whatever reason.

If the chart shows a strange series of reversions all at the same position for the greater number of features, then the chronology is faulty. Further investigation will reveal that it was some other feature that had reverted back instead.

Remember, it is the number of declining steps and overlaps in the chart that establishes the degree of accuracy for the chronology, and that the more features that can be included, the more certainty there can be. If enough features change enough times, the chance of this being a random arrangement is negligible.

Here is a screen shot of my flow chart for Series X. Along with divisions for the design groups, are tallies of the number of dies in each evolutionary step.

Screen shot of flow chart
Click on image for page discussing Series X in greater detail, including larger version of this chart.
6 - Interpretation of the Data

Once the chart has been constructed, certain patterns will be apparent. It might happen, that previous classes given for the coinage are too many or too few, or that their divisions are based on only one or two features. It might be seen that there are important breaks in the chronology that signify different issues. This is sometimes so obvious that separate flow charts have to be drawn, as the features themselves change so radically.

If the previous classification is now obsolete, all of the evidence of distribution patterns, declines in weight and precious metal content, coin to die ratios, and claims of die links between issues must be looked into. In my work on the Coriosolite series, I found that a few coins had been linked to the wrong part of the chronology because of superficial similarities. I was also able to make sense out of the distribution patterns where none existed before, and ended up reattributing some coins to a different tribe.

Certain observations about the symbology of the coins might become apparent. I found that similar positions on the coins tied in certain symbols, and that these symbols were not only unique to the location on the coin, but differing symbols could be seen to have the same meanings. This is perhaps the most difficult part of the research as it requires considerable investigation into the mythologies of the area and also areas that might have had contact and influence. It is the study of syncretism that holds the key to these interpretations, and if done properly, strong arguments can made for interpretations that without the evidence of the coins, would be impossible to state.

In the art-historical context; a demand for originality on the part of the die engravers became apparent; one of them incorporated this into his working methods, and it was only visible by his abandonment of masterpieces of design to embark on new experiments. Another die engraver achieved originality by arbitrary changes, all of which were borrowed from other artists, but used in no purposeful way.

Another die engraver loved to play with visual alliteration, and was so proud of this that he marked the dies especially so that they could be gathered together. Unfortunately, the moneyer mixed the obverses and reverses slightly so that his tags did not start at the time they should have.

As each die engraver is different, I cannot tell you how to interpret any flow chart for its artistic content, that is something that will have to be newly discovered for each series. Celtic die engravers had special training, and a set of aesthetic rules that were tied into their religious beliefs so thoroughly, that it is often impossible to separate the two, but the most neglected fact in Celtic numismatics is that the dies were cut by real people, like you and me, and that each of these people had their own personalities, weaknesses and strengths. Top