It is a remarkable fact that every relatively large piece of hollow ware, issued by the famous French silversmith foundry Christofle in 1846-1930, possesses a so-called numbering, i.e., each item is marked by its own ordinal number (Literature: 1). No other company in the world made such marking. Indeed, Christofle had even two numberings. The first one was used from c.1846 until 1862 (Literature: 2) (or, according to (Literature: 3, 4), until 1865) and was running from 1 up to at least 337243, which is engraved on the bottom of a small Christofle coffee-pot from my collection (Fig.1).

Fig.1: A coffee-pot, issued c.1862, and its ordinal number

After the change of the Christofle mark in 1862/1865 the second numeration started, that continued until a number which was very near to 3000000 (three millions). I have in my collection a tea-pot with the inscription "Grand Hotel du Louvre" and the ordinal number 2993546, see Fig.2.

Fig.2: A teapot with the inscription "Grand Hotel du Louvre", issued c.1930, and its ordinal number

In the last few years I was permanently trying to correlate the Christofle ordinal numbers to the real years of issue. The main problem was the lack of a sufficient number of the reference items containing the date of production. Now this problem is solved as I have enough reference items which allowed me to build a mathematical model connecting the production year and the ordinal number.

In 2006 I was lucky to purchase a couple of dated milk cans, bearing the numbers 1774104 and 1776073, both signed by the inscription "Salzburg 98" (Fig.3). I supposed that these milk cans with a number 1.77 million (1.77 M) were made in 1898 or a little earlier. Some of my colleagues argued that the inscription "Salzburg 98" might be "a model name and number". I did not agree with that comment, as French people of that period would never name their hollow ware model by the name of an Austro-Hungarian (or Bavarian) city. There are some Christofle models, named after French places like "Malmaison" or "Marly", this is possible, but certainly not "Salzburg". Later, I saw in Internet an auction for the silver-plated trophy produced for the Arles Regional Agriculture Competition organized by the Ministry of Agriculture (Ministere de L’Agriculture Concours Regional Agricole d’Arles) in 1899; that trophy was bearing the ordinal number of 1818091 (Fig.4), thus confirming the dating of my "Salzburg 98" milk cans.

Fig.3: A milk can with the inscription "Salzburg 98" and its ordinal number

Fig.4: An agricultural trophy with four turkeys made in 1899 and its ordinal number

Very recently, I bought two dishes with the numbers 2943686 and 2943692, both dishes have the same inscription "Albion Palais des Alpes 1928" (Fig.5).

Fig.5: A dish with the inscription "Albion Palais des Alpes 1928" and its ordinal number

The following important information on this matter was kindly given me by Professor Marc de Ferrière le Vayer (Literature: 5). The object with the number 1370632 was produced between December 1887 and January 1889; the item with the number 1502505 was produced between December 1890 and December 1891; the item with the number 1776073 was issued between April 1898 and February 1899.

With all this data it was possible to plot a graph and fit it with a quadratic polynomial function (Fig.6). This graph gives the dependence for the second numbering (Y) versus time period in years since the starting point (X) together with the approximating quadratic function. For the starting point either 1862 or 1865 could be used, however, the plot for 1862 showed better correlation between the experimental points while fitting and therefore was chosen as a final one. The time period (X) was calculated since the end of June 1862 (middle point in the year 1862). Everybody with basic algebra knowledge can easily calculate the corresponding numbering for any year between 1846 and 1930 using the expression given in Fig.6.

It should be emphasized that the smooth curve on Fig.6 is only an approximation to the real dependence. Certainly, during the long history of the Christofle foundry there were periods with higher production efficiency (e.g., due to the opening of new branches) or periods of decline (e.g., due to the war). Nevertheless, this graph is still quite useful.

Fig.6: Dependence of the ordinal number on the time period since the middle of 1862

In Table 1 the results for the reverse problem, namely, how to calculate a production year (how to date the item with the second Christofle mark), starting from the number on the object, are given. The total uncertainty (accuracy) in the year determination is about 3 years or less.

Table 1: Correlation between the ordinal number and the production time between 1862 and 1930

Now we will turn our attention to the first numbering which started somewhere after 1844 (I have in my collection a piece with first mark without any numeration) and finished in the middle of 1862 (as we have just established). According to the information kindly supplied by Prof. Marc de Ferrière (Literature: 5), the Christofle Museum purchased a set of hollow ware made by Christofle for king Louis-Philippe and in this set each piece had its ordinal number starting with 1. It is known that in 1847, Charles Christofle was appointed Fournisseur Officiel du Roi Louis-Philippe (Official Purveyor of the king) and, therefore, it is logical to suggest that the first numbering started sometime in 1846. Another priceless piece of information, obtained from Marc de Ferrière, was the dating of one hollow ware item with the first Christofle mark: the object with the number 149240 was produced between December 1857 and March 1859.

Using all this data, I again plotted a graph and fit it with a quadratic polynomial function (Fig.7). This graph gives the dependence for the first numbering (Y) versus time period in years since the starting point (X) together with the approximating quadratic function. The time period (X) was calculated since the end of June 1846 (middle point inside the year 1846).

Fig.7: Dependence of the ordinal number on the time period between the middle of 1846 and the middle of 1862

In Table 2 the results for the reverse problem, namely, how to calculate a production year (how to date the item with the first Christofle mark), starting from the number on the object, are given. The total uncertainty (accuracy) of year determination is about 3 years or less.

Table 2: Correlation between the ordinal number and the production time between 1846 and 1862

The author is indebted to Professor Marc de Ferrière le Vayer, former director of the Christofle Museum in Saint-Denis, Paris, for the priceless information and invaluable discussions. The author is also thankful to André Van den Kerkhove, Honorary Director of the Historical Museum of the Byloke of the City of Ghent, Belgium, for his courtesy.

LITERATURE
1. David N. Nikogosyan, An unknown mark for silver-plated hollow ware used by Christofle in 1930-1935, http://www.ascasonline.org/articoloGIUGN130.html
2. Marc de Ferrière, Christofle: 150 ans d’Art et de Rève. Dossier de l’Art, No.2, pp.3-73 (1991), [in French].
3. André Van den Kerkhove, private communication.
4. Walter van Dievoet, Dictionnaire des Orfèvres et Bijoutiers de Bruxelles et des arrondissements de Nivelles et de Hal-Vilvorde au XIXe siècle, Louvain, 2003, pp.1-485, [in French].
5. Marc de Ferrière le Vayer, private communication.