Skip to content

The 1870 census

Elissa Brown points us to this reproduction of the 56 beautiful pages of the Statistical Atlas of the Ninth Census, published in 1874. Takes forever to download it but it’s worth it.

Here’s a bit from page 38 (chosen for its humor value, but also because it’s pretty):


Page 46 is nice too.

Some of the designs are pretty good, some not so much. I’m not going to give lots of examples–hey, these guys were working back in 1870 and had to do everything by hand! I just wanted to point out that, no, these old graphs are not perfect–many of them could be improved upon in obvious ways–just to avoid the implication that these represent some sort of perfection. They represent an impressive level of effort and also remind us how far we’ve come.

My own standards have changed too. I remember twenty or thirty or forty years ago, spending many hours engrossed in the tabular displays of first, the World Almanac and then, the Statistical Abstract, just soaking up the numbers. Now I’d just be thinking about how these should all be graphs, but at the time I got a lot out of reading the numbers (the proverbial 10,000 hours, I suppose).

It’s been said that if there had been no such thing as basketball, Larry Bird would would’ve been a tall farmer. If there had been no such thing as statistics, I’d probably have become a mediocre mathematician.


  1. zbicyclist says:

    I initially intended to send this as a joke to a co-worker who asked me where he could find a time series of census data by age (he wasn't looking this far back).

    But this is a fascinating graph. One thing I don't understand is the pattern for Massachusetts women under 20 — the graph should look pyramidal, but theirs is chopped off. I see the same is even more true for neighboring Rhode Island and New Hampshire. Men show a similar but less pronounced pattern.

    Anybody remember what I've forgotten / never knew about New England during this time?

  2. Noumenon says:

    No idea what those graphs are measuring or represent. Didn't they believe in labels back then?

  3. ziel says:

    A plausible explanation is that the negative impact of modernity on # of children per household was already beginning to take shape in New England. That Massachusetts has a sharper truncation of female children might be just a statistical fluke. It does seem the more Northeast you get the more of a "Hershey's Kisses" shape the graphs take on, vs the more pyramidal shape in other states.

  4. zbicyclist says:

    The graph is labeled if you look at the full version, but the conventions are a bit different. In each case, there is a normalization to 1000 people (like percents norm to 100 people). So, for the Chinese graph, there are 1000 Chinese in it.

    Each level represents a decade of age. The lowest level is the first decade of life, the second level is teenagers, etc. So Chinese are dominated by men in their 20's (here to work on the railroads, if I remember that part of history).