Specific Gravity

Summary of discussion on Numism-l List from Archive from 98-01 to 99-05

From: Dr Gyula Petranyi 
Date: 1998 Apr 23 00:36:09 +0300

Specific gravity can be used as a simple measurement to find out the ratio
between the components of a bimetallic alloy (remember Archimedes and
heureka). The method has been used for coins, too, but I wonder if striking
changes specific gravity. Hammering can already make the flans more dense
and then striking the coin may add to this. Is this effect big enough to
make specific gravity measurement unreliable in numismatic research? Does
anyone know more about this?

From: Jxrgen Sxmod Date: 1998 Apr 22 23:33:20 -0700 I am often using specific gravity as a non-destructive testing of an alloy. Gold, which has 19.3 can by striking be highened to 19.6. I don't know the changings for other metals. But this is not the main problem. Impurities and gas bobbles can really change the specific gravity. So, I remember many years ago a beginner had bought a genuine minor 17th century silver coin and now tested the specific gravity, which in this case was only 7.x or 8.x and he thought it was a forgery. It gave a lot of troubles for some people, who thought they were experts! - And guess how my flaming was against them.
From: David Garstang Date: 1998 Apr 22 20:10:56 -0700 Specific gravity won't change due to striking solid metal with no bubbles. Metal is not compressable. It is malleable -- it will "squish" into a new shape -- but the volume won't be altered unless it contains gas bubbles. It seems to me that the striking would tend to squeeze out the bubbles, making it closer to the correct specific gravity reading. My question is, how do you get accurate enough volume and weight readings on an object that small? What type of equipment do you use, where do you buy it, and is it horrendously expensive? If not, I'd like to create my own setup.
From: Ibpanda Date: 1998, Apr 1998 18:52:17 EDT Well, the method of Archimedes will still work. You need a graduated cylinder of adequate diameter, callibrated in milliliters or finer. For tetradrachms that might be hard to find -- standard lab glassware would fit a denarius or an obol, but not a big old honking sestertius. If I pour 20 ml. of distilled water into the cylinder and then put in the coin and read the resulting volume as 21.6 ml I know the coin has displaced so many ml of water. Now the book says the density of pure silver is 10.49 grams per cc. (1 cc ~= 1 ml at 20 deg. C) So a nearly pure silver tetradrachm that weighs 17 grams should displace about 1.6 ml, unless I did the math wrong. A good laboratory supply catalog should have balances accurate to a hundredth of a gram -- not sure what they cost -- probably a few hundred US$. Most jewelers only use balances accurate to a tenth of a gram. I have no idea how analytical laboratories measure the volume of very small objects, but I imagine it's just some refinement of the displacement method.
From: Martins, Ken/SCO Date: 1998 Apr 23 17:21:50 -0600 Expanding on Ibpanda's suggestion to use liquid displacement to measure the volume of the coin: I would suggest adding a small amount of surfactant (soap) to help remove air bubbles attached to the coin. There is still a big issue of accounting for thick patina. This method may not be suitable for heavy patinated coins.
From: Warren Esty Date: 1998 Apr 23 18:58:26 -0600 Contrary to a recent suggestion, I do not think "the displacement method" commonly used for finding the specific gravity of coins directly measures the volume of the liquid displaced. Weigh the coin with a balance (to .001 gram if possible). Then weigh it again immersed in a dense liquid while it is suspended by a very fine thread. The displacement will cause it to weigh less by the weight of the liquid displaced (I think carbon tetrachloride used to be used because it is much denser than water and so gives a greater weight difference and thus more accuracy, but I don't know what is used now, given the hazzards of chemicals). Then the density of the liquid can be used to determine the volume of the coin from the difference in weights in air and liquid, and, with its weight in air, the density of the coin follows.
From: J. Mike DeWeever Date: 1998 Apr 23 22:28:34 EDT A "grain" scale used by all firearms cartridge reloaders, will accurately measure to +/- 1/10th (.1) of a grain, which is 1/ 70,000th of a pound (1 lb.= 7,000 grains ) 1 / 154th a gram (1 gram = 15.432 grains) these start at about $45.00 with a top end of 450 grains, there are also balance and electronic models which may go higher ... weight and price, call a local sporting goods store that sells firearms (you will be discussing "reloading scales") if they dont have one on the shelf they will gladly order one for you, Most are balance type so you could remove the "pan" and adapt for a "string and immersion" type situation. one mfr would be R.C.B.S. Just a quick thought as I am familer with this type scale and these are somewhat inexpensive for the degree of accuracy.
From: Peter Gaspar Date: 1998 Apr 24 09:00:55 -0600 (CST) Warren Esty is completely correct in his description of the method and his statement that it is the weight of liquid displaced that is measured directly. Carbon tetrachloride is a dangerous carcinogen and should not be used. Other perhalocarbons are not as dangerous but still toxic. The trouble is that to use them safely requires a well-ventilated room, but air currents disturb the weighing. I use water despite its lower density and higher surface tension, the latter a source of systematic error in the weighing. The advantage of water is that its density is known to many significant figures as a function of temperature. Since the density of a liquid changes enough over a few degree temperature range to affect the final specific gravity of a coin by >0.01 s.g. units, the temperature of the liquid must be measured and the density of the bath liquid at that temperature known. With a semimicro balance that weighs to 0.00001 gram, and using water, my specific gravity determinations of silver coins has an estimated error of 0.02 grams/cc. Each weighing must be repeated at least five times and the results averaged to achieve that level of accuracy.
From: Tony Clayton Date: 1998 Apr 27 22:43:30 GMT As a metallurgist by training, I would doubt very much whether striking a coin would significantly change the specific gravity (or relative density, to use a more modern term) unless the struck material was porous in the first place. A refence book I have regarding cold worked nickel indicates a density change of only 0.05%, which is recovered on annealing. Someone mentioned implosions. In these cases the pressure is exceedingly high for a short period of time, and this will compress the crystal structure significantly, as used in nuclear devices. However, when the pressure is removed the interatomic spacing would revert to normal. The biggest problem in determining composition by density is measuring the volume of the coin. The upthrust on a typical denarius of volume approx 0.2 cu cm is 0.2 grams when immersed in water. The presence of even a small bubble attached to the coin in the crevice of the inscription would make a significant error in the calculation, especially compared with the density differences as the silver content changes. The bulk modulus of gold is about 1.6E11 N/sq m if my calculations are correct, which means that the volume drops by 1% if the force exerted on a 1 sq cm coin is 16 tonnes, but this is elastic and thus reverts on the removal of the force. (My calculations are based on a Young modulus of 11.4E6 lb/sq in and a Poisson's ratio of 0.42 for gold, and were carried out well past bedtime, so I cannot be certain of their accuracy right now!)
From: James A. Schell Date: 1998 Oct 31 14:15:45 -0600 I am attempting to calculate the composition of certain coins by determining the density/specific gravity using a Mettler H70 analytic balance to measure the mass/weight in air and the mass/weight in water. I have encountered some problems. The Mettler H70 is readable to 0.1 mg. By calibration with precision masses, I am confident that the measurements are accurate to within 0.4 mg. My test coin is 1/4 troy ounce American Eagle bullion coin with the following composition: Au 91.67%; Ag 3.00%; Cu 5.33%. Using SG Au = 19.42 g/ml, SG Ag = 10.50 g/ml, SG Cu = 8.92 g/ml, and SG water(distilled and degassed) = 1.00 g/ml, adjusting to 22 Celsius, and assuming the ratio of Ag:Cu is 300:533, my measurements suggest a gold content of 90.72% (range:90.55% - 90.86%). If the only source of error is the balance, I should get a value in the range of 91.63% - 91.71%. Assuming the composition is as stated and reversing the calculation, the measured weight in water would seem to be 4.4 mg low. I added a drop of purified detergent to the water to reduce surface tension with no change in results. A second test using a sterling silver replica, yielded a composition of Ag 91.89% and Cu 8.11% (versus Ag 92.5% and Cu 7.5%). Again, the weight in water would seem to be 1.8 mg low. Does anyone have an explanation for the discrepancies, assuming, of course, that the compositions are correct?
From: Paul Date: 1998 Nov 1 13:43:19 +0000 Do check for other sources of error : You could always repeat the experiment using pure ethyl alcohol in place of water, which should reduce the surface tension problem. If you still get the same result, you can begin to assume that the alloy is wrong, not the method or the weighing. The reason that I say this is that the mix of metals within an ingot of an alloy is anything but homogenous. Although the mix within the alloy may be good whilst the temperature is high, there will be crystallisation as the ingot cools, and depending on the solubility of one metal within the other - and in the solution of the alloy, selective crystalisation will occur, the amount depending on the temperature, so pockets of high and low content, of one, or more of the constituents can and will occur. It will vary most towards the top centre of the ingot which is one of the last places to cool and is further complicated because of the presence of dissolved gases and other light impurities. The British Royal Mint did some measurements on silver ingots, back in the 1870-90 period. They are published in a Royal Mint Annual Report. It is more than a few years since I read the article, but as far as I can remember, the variation could be anything up to 7.5% off from the average, depending on the position from which the sample was taken within the ingot. Providing the your lab work is clean, which it is if you have checked your weighings to within 0.4 mg, I would assume that the error lies in the alloy rather than your technique. I have just checked our library and cannot find the article in question - so we must have it somewhere as an offprint - or it may have been reprinted elsewhere; it is certainly not with our Mint Annual Reports. I suggest that you try writing to one of the US mints, who will certainly know about the subject of homogeneity within of ingots of alloys. If that fails the try the Librarian and Museum Keeper of the British Royal Mint, G Dyer, who will almost certainly know the paper.
From: James A. Schell Date: 1998 Nov 1 15:15:39 -0600 Thanks for the informative reply. Several respondents indicated that my assumption that the SG of an alloy would be the proportionate algebraic combination of the SGs of the components was incorrect. I am now in search of empiric SG v. composition data.