Bulletin 21, 1973

Home
Français
Introduction
History
Annual Index
Author & Subject
Credits
Contact

The Compositional Analysis of 
French-Canadian Church Silver

by R. M. Myers and J. F. Hanlan
Canadian Conservation Institute*

Résumé en français

Pages  1  |  2  |  3  |  4

Introduction

*The authors would like to express their appreciation for the advice, co-operation, and guidance of Jean Trudel, and for the interest and encouragement shown by Dr Nathan Stolow throughout this work.

The traditional study of cultural artifacts by means of historical research and connoisseurship is becoming increasingly complemented by scientific examination and analysis.  Frequently the historian or curator may have new lines of research suggested to him by such scientific studies or have subjective appreciations validated. Often technical data will necessitate re-evaluation of an object.

In conjunction with research into materials for an important exhibition of French-Canadian and French silver - Silver in New Francet - held at the National Gallery in 1974, the chemical analysis of Church silver currently in the National Gallery collection was undertaken. It was felt that the establishment of composition would provide important information on aspects of silver and silversmithing concerning work - shop practices, availability of raw material, and possible compliance with French guild rules in early Canada. The emergence of correlations between composition and silversmith (or period) was also a possible result, as was the discovery of repairs, additions. or mismatches. Comparison of our results with those from a similar study of early American silver currently underway at the laboratory of the Henry Francis du Pont Winterthur Museum in Winterthur, Delaware (1) can be expected to be illuminating.

Church silver has certain unique advantages for a study of this type, because in general the silversmith used the best grade of silver available - and the pieces are in good condition, positively identified, and well-polished. The National Gallery's particular collection was also convenient in that the main artists represented - François Ranvoyzé, Laurent Amiot, and François Sasseville - did most of their work from the years 1760 to 1864, in three different periods, including one of the most prosperous eras of silversmithing in Canada.

One major problem with a study of this type is the fact that no original Canadian silver was used because there were no mines discovered, and thus all articles were produced from remelted silver objects, coins, and scrap. Out of the melting-pot came a whole range of alloys; no worker can therefore be expected to exhibit any consistent fineness of metal, or even the constant use of metal from one particular source.

Experimental

All analyses were done using X-ray fluorescence spectroscopy with an energy dispersive detector (EDX) (fig. I). Because of the great efficiency of this type of detector, compact and safe radioisotope excitation can be used, and analyses are performed very rapidly. The main advantage, however, for museum applications is that simple elemental analyses can be done on select areas of a precious or fragile object without the need for removal of samples. (For a further discussion of this technique, see Appendix, and notes 2, 3, 4 and 5.)

A calibration curve was prepared by positioning a series of copper (Cu) and silver (Ag) standards (% Cu / % Ag - 50 / 50, 20 / 80, 10 / 90) in front of the detector and counting the Cu Kx and Ag Kx characteristic X-rays and two background channels around the copper peak, which were subtracted. Background in the vicinity of the silver peak is insignificant and was ignored. The corrected copper to silver counts ratio was then calculated and plotted as a function of % Cu / % Ag on log-log paper according to a method used by Bertin, (6) resulting in a straight line plot. By using a ratio method, the effects of instrument instability, radioactive source decay, and distance or orientation from sample to detector are minimized, and constant recalibration is unnecessary. The log-log plot was also necessary to ensure that extrapolation of the curve to the 90-100% Ag region could be done with confidence. To check the curve during the study a silver blank of approximately 80% Ag was polished until both sides gave the same copper/silver counts ratio, and then three pieces cut from it underwent wet analysis, giving 79.0% Ag. Over a two-month period, with readings on twenty-three different days, the EDX gave an average reading of 79.0 + / -.5% Ag at the 95% confidence level for the remaining portion of the blank.

A silver object was analyzed by positioning it so that the area of interest was about one-eighth of an inch in front of the excitation Source, and the total spectrum count rate was then adjusted to about 2,000 counts per second using the collimator. Two readings for each area were taken in order to improve the statistical accuracy. From this, the copper / silver ratios were calculated and the percentage silver was read from the calibration graph. Finally the two percentages were averaged.

Because any X-ray fluorescence method only analyzes the surface (e.g., the Cu Kx fluorescent X-rays originate from a maximum depth of .03 mm in a 95% Ag sample), it is most important that the surface be representative of the composition of the object as a whole. This is not always the case, because in almost all alloys the effect of surface enrichment, or more correctly, surface depletion, of one or more components changes the surface composition with respect to the interior. The effect can be caused by oxidation, or by leaching out of a component, or simply by diffusion of one component through the matrix. The enrichment is not always apparent, because in general no pitting, corrosion, or any other visible change takes place. Objects fabricated from silver were also typically whitened by a process known as blanching, which leaves a silver-rich layer on the surface. It is important then that these effects do not cause misleadingly high results for the silver percentage. In a study by Carter (7) in which the enriched surface layer on silver coins was removed by blasting with aluminum oxide powder, it was found that usually about one-thousandth of an inch (0.0254 mm) of the surface had to be removed to get an accurate X-ray fluorescence reading. Removal of just 0.014 mm of surface metal from a silver coffee-pot analyzed by EDX by Hanson resulted in a jump from 2.2% to 5.8% Cu. It was thus decided that with fairly constant polishing on a smooth surface, no enriched layer would exist.

Next Page |  Results

1  |  2  |  3  |  4

Top of this page