Modeling Digression: Coin Weights


Measuring a euro coin
Too much philosophy, let’s get some data!

Part of the motivation for the recent posts on modeling is that I’m writing a paper for a friend. It takes off from the hexnut weight data and cubic model that appear in The (still unpublished, sigh) Model Shop. More on that soon, because I’ve found some interesting features in hexnut data. But first, another friend sent me a link to this post on doghousediaries I now share with you. Check it out.

Because I had just been doing hexnuts, I immediately thought about coins as an alternative: how do you suppose the weight of coins (Physics people: I will be reporting weight in grams. We both know that’s mass, not weight. We can handle that, right?) is related to their size?

I mean, hexnuts look more or less geometrically similar (they’re not, but that’s the other story), so you’d think their mass is cubic in the linear size: a nut that fits a half-inch bolt ought to be eight times the weight of one that fits a quarter-inch bolt. And that model fits pretty well.

So as teachers, we look for situations that stretch that understanding. EGADs has you cut out cardboard squares or circles and weigh them; since the difference in size is two-dimensional, the weight-size relationship is quadratic.

But what about coins? They look kind of two-dimensional, but the big ones tend to be thicker. So are they cubic or quadratic or something in between? Well. Whenever I come back from some other country, I usually have leftover coinage, which gets saved in that jar in the top drawer. And I have a good scale and (I know, I’m a measurement geek) a decent micrometer. So it’s an empirical question.

Yet we expect trouble—or at least more departures from any model—from the outset:

  • The coins are made of different metal, so should have different densities.
  • It’s obvious, looking at a bunch of coins from different countries, that although the big ones are generally thicker, the relationship between thickness and diamater is not directly proportional.

I would ordinarily just let the noble reader measure his or her own coins, but I have just done this and know that it’s a pain in the patootie. So without further ado, some data!


Coin weights with a cubic model

Coin weights with a quadratic model
Cubic model Quadratic model

I did record the overall color of the coins (silver-ish, gold-ish, mixed) as a poor proxy for density, and there’s no obvious pattern. In the graphs, the four red points are Mexican coins (1, 2, 10, and 20 pesos). I selected them because their overall design is so unified and because they are generally so impressive and heavy. The 20-peso coin, man, you could wear it around you neck and feel like you’ve won a race or something.

Anyhow, it looks as if—despite all the variation—the cubic is plausible. The quadratic, however, is not. The trend is just too steep; the coins are not fundamentally two-dimensional objects.

Well. We have to look at a linearized plot, don’t we?

Sure we do. But I’ve deprived you of enough fun. Here are the data so far. I am still measuring thicknesses, so I do not include it yet. Mass is in grams, diameter in inches. It’s tab-delimited. When I copy from the page and paste into a spreadsheet (or a Fathom collection) it works right:


number	unit	country	mass	diameter	color
25	cent	USA	5.68	0.9526	silver
25	cent	USA	5.61	0.9538	silver
25	cent	USA	5.66	0.9518	silver
1	dollar	USA	8.03	1.0444	gold
1	dollar	USA	7.99	1.0446	gold
1	dollar	USA	7.95	1.0432	gold
10	peso	MEX	10.23	1.08268	mixed
1	krone	SWE	6.97	0.9843	silver
2	pence	GBR	7.03	1.0221	gold
20	peso	MEX	15.83	1.25984	mixed
5	rand	RSA	6.93	1.0203	silver
1	real	BRA	6.83	1.0613	mixed
50	centavo	BRA	3.96	0.9036	silver
2	peso	MEX	5.17	0.9092	mixed
5	centavo	BRA	4.11	0.869	gold
5	cent	USA	4.95	0.8339	silver
5	cent	USA	4.92	0.8362	silver
5	cent	USA	4.98	0.834	silver
2	euro	FRA	8.55	1.00394	mixed
1	euro	GER	7.51	0.9144	mixed
1	pound	GBR	9.55	0.9871	gold
10	centavos	BRA	3.6	0.8656	silver
1	peso	MEX	3.92	0.8293	mixed
5	euro-cent	EUR	3.94	0.8367	gold
1	cent	USA	2.46	0.7507	gold
1	cent	USA	2.49	0.7487	gold
10	cent	USA	2.25	0.7041	silver
1	krone	DEN	3.68	0.7974	silver
10	öre	SWE	1.45	0.5904	silver
1	dollar	CAN	6.96	1.02362	gold
10	krone	NOR	8.97	0.9472	silver
About these ads

About Tim Erickson

Math-science ed freelancer and sometime high-school stats teacher
This entry was posted in content, curriculum development, modeling and tagged , , , , , . Bookmark the permalink.

3 Responses to Modeling Digression: Coin Weights

  1. A.K. says:

    Did you have any aluminum coins in that jar? They’re so light that I imagine they would seriously throw things off.

    • Tim Erickson says:

      Good point! But no, I don’t. I now have the thicknesses, and although they vary a lot, they let me at least get a slighty-better-than-ballpark (perhaps infield?) estimate of the densities of the coins. Those numbers came out between about 6 and 9 (g/cc), not the 2.7 we would get for Al.

  2. Interestingly, I got a better fit using a mass=A*diameter^3 than A*diameter^B. That is, the improvement in fit from letting B vary was less than the cost of increasing the number of degrees of freedom.

    I suspect that putting in Al coins or Au coins would through things off a bit. Just about all the standard coins these days are mostly zinc, with small amounts of copper or nickel. Your denser ones probably have more copper (silver is possible, but unlikely these days).

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s