I look forward to any comments, additions or corrections to this post.
Observation of some Kan’ei Tsūhō Namisen Attributes
fairfield, January 20, 2025
- Summary
Conventional numismatic wisdom teaches that 11 wave Kan’ei Tsūhō Namisen of 4 mon value from the Meiwa period N#162214 appear “brassy”, Bunsei period N#156603(with some lead addition) have a reddish tint, and Ansei period pieces N#162216(with a high lead content) have a blackish appearance. The color is generally assumed to be caused by changes in the alloy composition; or, to trace contaminant(s) in the starting raw metals. With this backdrop, eight 11 wave Namisen were judged for color, measured for density and magnetism, and semi-quantitatively evaluated for major elements in the composition. The results suggest that coin composition varied over a wide range and that color may not be a reliable indicator of casting period. On the other hand, lead content and, in the absence of chemical analysis capability, the coin density, may be more suitable; however, in the latter case, the wide range of observed density values may make confident identification of some coins impossible.
- The Target Composition of 4 mon Kan’ei Tsūhō Namisen
In his 1891 paper "Abridged History of the Coins of Japan," van de Polder gives us the following target compositions by casting period[1].
Period Copper Zinc Tin Lead
Meiwa (casting 1769-1788) 68 24 8 0
Bunsei (casting 1821-1825) 75 15 0 10
Ansei (casting 1857-1859) 65 15 0 20
This information has been repeated numerous times, for example in Hartill's 2011 text "Early Japanese Coins"[2] and Robert Jones 2007 text “A History and Guide to the Copper Cash Coinage of Japan”[3].
However, the 2020 JNDA Japanese Currency Collection Guide gives the composition of Meiwa period coins as 68% of copper, 24% of zinc, and 8% of lead, that is, Van de Polder's 8% tin is replaced by 8% lead[4]. Potentially important is that the Bunsei and Ansei compositions match, thus, with each advancing casting period, the target lead content was increased up to 20 weight percent.
- Predicted Density as a Function of Composition
To first order, metal density is straight-forward to predict from composition using the density and weights (thus the volumes) of the starting materials. Using both Meiwa compositions we have the predicted densities for all three periods given in the table below.
Period Expected Color Target Composition Density [gm/cm3]
Meiwa Brassy 68 Copper 24 Zinc 8 Tin 0 Lead 8.30
Meiwa Brassy 68 Copper 24 Zinc 0 Tin 8 Lead 8.58
Bunsei Reddish 75 Copper 15 Zinc 0 Tin 10 Lead 8.81
Ansei Blackish 65 Copper 15 Zinc 0 Tin 20 Lead 8.99
Not surprisingly, the predicted density increases with lead content. The “target” densities of Bunsei and Ansei pieces are only about 2% different, and any density measurements will need to be precise to better than 0.06gm/cm3 to confidently distinguish these two coins.
- Visual Inspection and Density of Eight 4 mon Kan’ei Tsūhō 11 Wave Namisen
Eight coins were randomly assigned letter designations A through H. Color was judged under sunlight illumination concentrating on both surfaces as similar to the method taught by Ogawa in his 1969 text "Shin Kan'ei Sen Kanshiki to Tebiki"[5]. The following language was used to denote color: Brassy, Less Brassy, Little Bit Dark, Little More Dark, Darkest, Reddish, and Blackish. Note that the first five terms are possibly all "brassy" coins in order of increasing patina on the surface. In the table below, Obv and Rev indicate the obverse and reverse faces, respectively.
Density was measured with a Mettler AG204 balance by the Archimedes method (weighing the sample first in air and then supported in DI water with 0.5% addition of Photo-Flo 200 at a known temperature to 0.0001gm) and is given below in gm/cm3. The density measurements were corrected for buoyancy of air and were replicated ten times for each coin. Standard deviations were between 0.005 and 0.009 gm/cm3. Without the addition of Photo-Flo wetting agent, surface tension and trapped bubbles on the samples increased the uncertainty to the range of 0.02 to 0.08 gm/cm3, near the limit to distinguish between Ansei and Bunsei coins of the target composition. The impact of Photo-Flo 200 addition on the density of DI water is evidently minimal as measurement of a density standard (NIST SRM 8041) agreed within 0.002gm/cm3.
Coin Obv Color Rev Color Assignment Based on
Obv Color Rev Color
A Reddish Reddish Bunsei Bunsei
B Brassy Little More Dark Meiwa Meiwa/Ansei
C Less Brassy Brassy Meiwa Meiwa
D Little Bit Dark Less Brassy Meiwa/Ansei Meiwa
E Little More Dark Little Bit Dark Meiwa/Ansei Meiwa/Ansei
F Darkest Darkest Meiwa/Ansei Meiwa/Ansei
G Reddish Reddish Bunsei Bunsei
H Blackish Blackish Ansei Ansei
Coin Obv Color Rev Color Weight Density Assignment Based on
[gm] [gm/cm3] Density
A Reddish Reddish 5.286 8.914 Ansei/Bunsei
B Brassy Little More Dark 4.858 8.243 Meiwa
C Less Brassy Brassy 4.602 8.193 Meiwa
D Little Bit Dark Less Brassy 5.088 8.095 Meiwa
E Little More Dark Little Bit Dark 4.677 7.782 Meiwa
F Darkest Darkest 5.173 8.316 Meiwa
G Reddish Reddish 3.992 8.688 Meiwa/Bunsei
H Blackish Blackish 5.678 7.907 Meiwa
Coins B through F are possibly all “brassy” with some tarnish; but, to be conservative, the color assignment for darker faces was made as "Meiwa/Ansei". On the other hand, A and G are distinctly reddish and coin H is definitely blackish (much darker than even F). But the key take away was that color and density suggested different assignments for coin A (reddish, should be Bunsei, but density is a little high, closer to Ansei assuming the target composition), coin G (again reddish, but now density is a little low, closer to Meiwa) and coin H (blackish, should be Ansei, but density is less than predicted for even a tin containing Meiwa coin). The wide density range, many much less than even the lowest predicted density of a Meiwa piece, also suggests a wide range of coin compositions!
- Chemical Analysis
All eight samples were submitted for chemical analysis by wavelength dispersive X-ray fluorescence (XRF) spectroscopy. Note that XRF only probes the sample chemistry at and just below the exposed surface. We also emphasize here that lacking a reference standard the results are only semi-quantitative, the values below are not the measured composition but rather just intensity values. However, one immediate finding was that none of the eight coins showed a strong signal for tin. Focusing only on the observed three major components of copper zinc and lead, the signal levels are given below.
Coin Obv Color Rev Color Copper Zinc Lead Assignment
Based on Lead
A Reddish Reddish 73.3 9.4 15.3 Ansei
B Brassy Little More Dark 72.3 21.1 1.6 Meiwa
C Less Brassy Brassy 80.2 18.7 1.1 Meiwa
D Little Bit Dark Less Brassy 79.6 19.5 0.9 Meiwa
E Little More Dark Little Bit Dark 81.6 16.0 2.4 Meiwa
F Darkest Darkest 84.6 14.8 0.6 Meiwa
G Reddish Reddish 82.8 10.0 7.3 Bunsei
H Blackish Blackish 85.2 13.9 1.0 Meiwa
Also detected at lower levels (but still 10 times over background) compared to copper and zinc were silicon (possibly carry over from the sand molds used to cast these coins as the analysis depth for this element is only 0.002mm), aluminum (actually assigned to the sample holder in the XRF instrument), iron, sulfur, phosphorous, magnesium, calcium, chlorine and alkali metals. Three distinct signal levels for lead are revealed in the results: 15.3, 7.3 and less than 2.4. The signal levels less than 2.4 were comparable with that of some of the other contaminants found in the spectroscopic data, and near absence of lead helps to account for many of the coins having lower than expected density values. Most importantly, the three lead signal level ranges again suggest assignments at issue with those based on color alone but more consistent with the density results, particularly for coin A and coin H.
- Other Links to Coin Color
To search for another cause of color variation two checks were completed. First, it was observed that some of the coins exhibited magnetism and it was considered if residual iron contamination might be a cause of color variation (after all, rust often has a reddish hue). Magnetism was evaluated as an “equivalent gm weight of force” (denoted by M below) by measuring the ability of a fixed magnet to both lift and translate each coin suspended by a thread under tension through an angle away from vertical. The angle, X, when the fixed magnet can no longer be attached and releases the coin gives M by M = m (sin X /cos X) where m is the measured weight of the coin in gm (thus factoring in and accounting for the different coin weights). This measurement was repeated at least five times using a different location on the coin edge for each trial and the standard deviation for all but coin F was less than 0.1 units. Coin F was highly anisotropic in its magnetic property, appeared to be more concentrated in (presumably) iron in one quadrant of the coin, and had a standard deviation of 0.5 units.
Second, as is well known in the metal community, brass color is often simply a function of composition even in the absence of impurities. We thus inspected the results for any correlation between bulk chemistry and perceived color. Below, the ratio values for lead to copper (Pb/Cu) and zinc to copper (Zn/Cu) multiplies by 100 are included.
Coin Obv Color Rev Color M [gm wt of force] 100xPb/Cu 100xZn/Cu
A Reddish Reddish 0.0 20.3 12.5
B Brassy Little More Dark 0.7 2.1 27.3
C Less Brassy Brassy 3.4 1.4 23.3
D Little Bit Dark Less Brassy 0.7 1.2 24.5
E Little More Dark Little Bit Dark 0.6 2.9 19.6
F Darkest Darkest 3.8 0.7 17.5
G Reddish Reddish 0.0 8.8 12.1
H Blackish Blackish 3.1 1.1 16.3
Key observations were that reddish coins show no magnetism but the blackish coin shows magnetism strength bracketed by the various non-blackish coins. We believe there is no link evident between iron and coin color. The XRF results for iron did not track with the measured magnetism, and we believe the latter is a better indicator of bulk iron content and the former subject to both surface effects (the analysis depth being less than 100um) and high uncertainty (as a point of fact, none of the contaminant levels observed by XRF correlated with coin color, but the expected high uncertainty for contaminants means this was not a rigorous check).
On the other hand, there appears to be a trend with both ratio values of lead and zinc to copper. In particular, a zinc to copper ratio less than 14 is reddish, a ratio between 14 and 20 are dark to blackish in color (with darker color for smaller ratio value), and a ratio value greater than 20 appears brassy to the eye. Again, these results are semi-quantitative, other equipment might give different ratio values, but the trends should be preserved.
For comparison, consider modern “red brass” and “yellow brass”. On May 21, 2022 from
https://www.rocheindustry.com/red-brass-vs-yellow-brass/ we have
Red brass is 88% copper, 8%-10% zinc, and 2%-4% tin and
Yellow brass is 60-70% copper, 30-40% of zinc, tin, and lead in traces[6].
This web site also states that red brass is associated with higher copper content and yellow brass takes its color from the higher zinc content. These statements are consistent with small Zn/Cu value being reddish and large Zn/Cu value being brassy (yellow).
What does this then tell us about assignment to the Meiwa, Bunsei and Ansei periods?
Period Expected Color Target Composition Target Zn/Cu Ratio x 100
Meiwa Brassy 68 Copper 24 Zinc 8 Tin 0 Lead 35.29
Meiwa Brassy 68 Copper 24 Zinc 0 Tin 8 Lead 35.29
Bunsei Reddish 75 Copper 15 Zinc 0 Tin 10 Lead 20.00
Ansei Blackish 65 Copper 15 Zinc 0 Tin 20 Lead 23.08
The values above suggest the target compositions (independent of lead or tin in Meiwa coins) follow the same pattern between predicted color and target Cu/Zn ratio (remember, the XRF chemistry values are all semi-quantitative, so only watch for trends, not quantitative agreement with prediction). So, the target compositions might be expected to give the reported colors. But, in the case of the eight coins discussed here, the actual retained metal contents in the coins are evidently far from the published target levels.
- One Other Possibility
The four mon coin contained only a little more copper than a one mon coin. Thus, it was an attractive target for illegal coin casting by the various clans across Edo period Japan (using 1 mon coins as feedstock). One would expect that the official 4 mon pieces would have enjoyed more accurate production methods and have density values close to target. If the density and chemistry variation reported here is from private casting operations, we might then propose that, since coins D, E and H have density much lower than the entire target range, they might be counterfeits of that time period. However, all the other coins have a density st least close to one of the target values, and we would still have to wonder at reddish coin A with density and chemistry matching most closely an Ansei coin but then having the wrong coloration. Either official or illegal, the implication is that a serious collector should consider attributes beyond color to make period assignments of Kan’ei Tsūhō Namisen.
- Conclusions
The retained chemistry of the Namisen in this study covered a wide range of compositions, many far from target based on density alone. If there was a gradual addition of lead over time from Meiwa to Bunsei to Ansei periods as reported elsewhere, the lead level may be a more reliable method to identify casting period. Although lead may be a contributing factor, variations in retained zinc and copper in the coins studied here appear to determine coin color INDEPENDENT of the assigned mint period based on color alone. Assignment to specific casting periods might then best be done by quantitative evaluation of lead content. This might also be done semi-quantitative as the case here; and, in the absence of chemical analysis capability, coin density might be a surrogate indicator. However, the wide variation in coin density may make specific assignment of some coins impossible by this property alone.
- A Final Remark
Some of this content was posted a few years ago on a blog in Japan maintained by kosenmaru. Subsequently, the quality of many of the measurements, particularly of density, were improved. Conclusions are the same but some of the data is of course changed.
- References
1. Van de Polder, Leon. "Abridged History of the Copper Coins of Japan." Transactions of the Asiatic Society of Japan, XIX (1891): 419-500.
2. Hartill, David. "Early Japanese Coins". Bright Pen. 2011.
3. Jones, Robert. “A History and Guide to the Copper Cash Coinage of Japan”. Kearney, NE: Morris Publishing. 2007.
4. “Nihon no Kahei Shūshū no Tebiki.” Japan Coin Dealers Association, 2020. Also found on May 10, 2022 at https://www.japanese-wiki-corpus.org/history/Kanei-tsuho.html and its original Japanese site https://www.japanese-wiki-corpus.org/jp/history/寛永通宝.html.
5. Ogawa, Yoshinori. "Shin Kan'ei Sen Kanshiki to Tebiki." Tokyo: Bankoku Kahei Yōkō. 1969. To be fair, Ogawa puts emphasis on the coin edge rather than the obverse/reverse faces.
6. "Red brass vs yellow brass: What are the Differences?." https://www.rocheindustry.com/red-brass-vs-yellow-brass/. Accessed May 21, 2022.