# Determining average number of people that have same surname as random person?

This website gives the approximate number of people with a particular surname in all countries: http://forebears.io/surnames

It spits a number, for example "1500 people", but the problem is that I don't know if that's high or low, because I don't have the average to compare it to...

If we picked a random person, what's the average number of people that have the same surname as that person?

Either for the whole world, for the U.S., for France, or for the U.K., any would be good enough.

It would be even better if you could also provide the median and the standard deviation.

And it would be even even better if you could provide a histogram representing the percent of people who carry a surname carried by between k*1000 and (k+1)*1000 people, i.e. with the x-axis representing bins “Surnames carried by between k*1000 and (k+1)*1000 people” and with the y-axis representing the percent of people.

• Isn't this just the same as asking how many surnames are in the population in question? I don't mean that the number is the same but the answer to the question is surely population / surname-count? Or is it so late I'm missing something? Nov 26, 2016 at 0:47
• @AdrianB38: No. Imagine there were 100 people, 99 of them with the surname “A”, 1 of them with the surname “B”. Your formula leads to a wrong answer: population / surname_count = 100/2 = 50. The correct formula would be: sum_on_all_surnames_s(number_of_people_with_surname_s^2) / population = (99^2 + 1^2) / 100 = 98.02 Nov 26, 2016 at 6:14
• Point taken. It clearly was late! Nov 26, 2016 at 7:14
• The service is great but the results are totally unreliable. I will explain. For example, I know at least the whole branch of Merbach's in Russia, but they are not represented in the service. Also the different spellings of surnames co-exists. Such a service will succeed only if every country will publish it's censuses and phone catalogues. But it is not possible. Particularly, if we speak about least developed countries Jul 24, 2017 at 14:53

My suggestion would be to investigate WolframAlpha (https://www.wolframalpha.com/) if you haven't already, at least for the United States. If you enter the term "surname xxxx," it will produce a number of relevant facts based on US Census data. For instance, "surname Jones" shows that based on the 2000 Census:

Jones is the 5th most common last name
1 in 198 people had it, for total of 1.363 million people
57.69% were white
37.73% black
1.85% mixed
1.44% Hispanic
etc.

It also lists notable people with that surname. Incidentally, you can do the same thing with given names (i.e. "given name John"), but that return shows two graphs, one that shows the percent of births each year having that name and one that shows the expected age distribution of people with that name. It also gives the etymology of the name.

I feel like I have done this in the past with WolframAlpha for names in the UK, but I can't figure out how at the moment.

There are some other example name based queries here: https://www.wolframalpha.com/examples/Names.html

Something else I just discovered is that if you type in an incomplete word with wildcards, it will attempt to complete the word, a la Wheel of Fortune. For instance, typing in "?ic?u?e" produces the possible words "bicause," "ricture" and "picture." [And yes, the first two are actual words, or in the case of bicause, a known variant spelling.] This could be a useful feature for those difficult to read census entries.