Celebrities Always Die in Threes

Three are some old sayings, such as "what comes around goes around", and "celebrities always die in threes" which are so obviously flawed that I wince when I hear them. However, I've found that it is best to occasionally test some of the things you already think you know, so you don't make an ass out of yourself.

On July 28th, I decided to keep track of all the deaths of famous people for August 2005.


To determine who could be considered "famous" for this study, I decided to use a 21-point popularity and success matrix include anyone who's death was worthy of the front page of Google News.

In August, exactly eleven famous people died: King Fahd, Little Milton, Robin Cook, Ibrahim Ferrer, Peter Jennings, Barbara Bel Geddes, John Johnson, Thomas Herrion, Robert Moog, Luc Ferrari and Brock Peters.

Many famous people did NOT die, including Clint Eastwood, Ringo Starr, Alex Trebek, Zsa Zsa Gabor, Regis Philbin and Debbie Harry.


Here is a timeline of their deaths. As you can see, none of them died in threes.
I figured those three-person groupings were just convenient triples that could be found in any set of data. To test this, I created a celebrity death month generator, which generated a month's worth of celebrity death dates.



The first test generated nine random dates: 20, 26, 12, 9, 26, 27, 25, 30 and 21st. 


There were not any grouped threes in this month.







The second refresh generated: 18, 8, 1, 20, 16, 21, 31, 31 and 20th.


Here again, there weren't really any grouped threes, although one was close, the 18, 20, 21 group was close, but the celebrity death on the 16th kind of made it a "dying in fours", and the gal dying on the sixteenth would blur the edge of that grouping into a "dying in fives".



The third refresh generated 29, 28, 12, 22, 6, 5, 16, 6 and 2nd.


Ah! Here there was a little grouping of three imaginary celebrity deaths. Occasionally, simulated celebrities did die in threes!


The fourth refresh generated: 7, 20, 4, 3, 18, 7, 4, 30 and 5th.

Two thirds of these nine were clustered into the first few days of the month. I could pick out three, but there were really six in there all grouped together. When this kind of grouping happens among real celebrities, I'm sure the three in a similar field, (i.e. music, television or political) become the celebrated three, while the other celebrity dead remain unmentioned.



The fifth batch was very well distributed!    1, 14, 27, 3, 10, 6, 13, 17 and 24th.


No triples, unless I drew a pretty big circle.

The sixth batch was generated, and when it was graphed, it looked like it might also could have a three. 7, 1, 14, 31, 12, 28, 16, 11 and 16th.

The fourteen and two sixteens  were clumped all together. I think a grouping like this would definitely generate some buzz about a "dying in threes" event.


Charting real deaths in August had not supported the dying in threes theory. It didn't happen.

I suspect, and the random death date simulator indicates that random occurrences can generate death dates which could easily be lumped into theory-supporting threes. I think the easiest way to summarize it is that if you are trying to see groups of three in a crowded month, you can almost always find them.


Contact Rob
science club | How much is inside? | Home | Incredible Stuff I Made
Last updated Sept 12th, 2005.
  • Photographic Height/Weight Chart
  • The Weight of Clothing
  • The Television Commercial Database
  • Terms and Conditions Copyright 2005 Cockeyed.com