#### Introduction

The Doomsday Algorithm can be used to mentally calculate the day of the week on which any date falls. With a little effort most people can learn to apply the algorithm mentally without pencil or paper. If you're interested, you can learn the method behind the Doomsday algorithm at rudy.ca or at Bob Goddard's site.My purpose is not to describe the details of the Doomsday algorithm, but to describe an improved method to implement part of it (calculating Doomsday for an arbitrary year) which I have discovered. Below, I'll describe the method first using mathematical notation, which unfortunately appears needlessly complicated, but then I'll show how to easily perform the calculations mentally without the need for memorizing cumbersome equations.

#### Background

First, a key component of using the Doomsday Algorithm is knowing how to calculate Doomsday for an arbitrary year, after which you can figure out the day of the week for any date in that year. The usual formula described to make this calculation can be written asint(y/12) + y mod 12 + int((y mod 12)/4)

where y is the last two digits of the year. The notation "int (a/b)" means to take only the integer part of a/b; "a mod b" means to divide a by b and take the remainder. For example, int (4/3) = 1, and 10 mod 7 = 3.

This result mod 7 is added to Doomsday for the base year of the century (e.g. 1900 or 2000) to get the final answer. For 1900, Doomsday is Wednesday (3, where Sunday is 0).

For example, Doomsday 1953 is

int(53/12) + 53 mod 12 + int ((53 mod 12)/4)

= 4 + 5 + int(5/4) = 4 + 5 + 1 = 10 mod 7 = 3

Adding 3 (Doomsday for 1900) to Wednesday gives Saturday.

This formula is not as bad as it looks, but it does require modular arithmetic, four divisions, and remembering three different numbers to add together. For me, this was always the most tedious part of applying the Doomsday algorithm.

#### An Easier Way!

Fortunately I have discovered an easier way to make the same calculation with much less effort!For a leap year only, my new simpler formula is

7 - (y/2) mod 7

where y is the last two digits of the year. This result is added to the Doomsday for the base century year as before.

For example, for 1980 Doomsday is

7 - (80/2) mod 7 = 7 - 40 mod 7 = 7 - 5 = 2.

Adding to Doomsday 1900, the final answer is 2 + Wednesday = Friday. While my formula is simpler than the original, it still involves some modular arithmetic and divisions. But there is a simple shortcut which makes applying the formula easy.

Here's the trick: To apply this in practice mentally, you would simply divide the leap year by 2 and subtract the result from the next largest multiple of 7. No divisions by 12 or 4, no modular arithmetic, and less memory effort is needed. If you can divide an even number by 2 and add and subtract, you can do it quickly and easily.

Even better: If the year is not a leap year, just repeatedly add 11 to the year until you get a leap year and apply the same formula. At most this will require adding 33 to the year.

#### Examples

Returning to the first example, Doomsday 1953 is the same as Doomsday 1953 + 11 = Doomsday 1964.Dividing 64 by 2, the result is 32 which subtracted from 35, the next multiple of 7 to give 3, the same as above. Added to Wednesday (Doomsday 1900) the result is again Saturday.

Another example: 1962 is not a leap year, so we add 22 to get 1984. 84/2 = 42, which is a multiple of 7 so the result is 0 and Doomsday is Wednesday. (In this case the original formula might be easier since 84 is a multiple of 12). If you're familiar with the traditional method given above and my easier method, it is easy to apply the simplest one for any case, which I believe is usually the newer formula given here. I only find the older method easier if the year is a multiple of 12. I especially like the new method for finding the Doomsday of an odd year, which I always hated to do using the older method.

Update: When the process of adding 11 takes you across a century boundary which is not a leap year (e.g. 1900 or 2100), the process is similar as pointed out by Robert Goddard in a comment to this post. As an example of crossing non leap year centuries, I'll use his example of 2095. We add 95+33=128. Since we crossed a non leap year (2100) use "128" instead of "28". We divide by 2 to get 64, and subtract from a big multiple of 7, to get 70-64=6. Added to Tuesday (Doomsday 2000, the century we started in) and the result is Monday. When crossing century years that are leap years, like 2000, the original method works without adjustment.

It is easy to verify this method of adding 11 works by examining the following table which I have borrowed from rudy.ca. In this table the leap years are bold. You can see that every leap year is the final year in a chain of 4 years which differ by 11 and which have the same Doomsday. Every non leap year is a member of such a chain. Because it is easy to find Doomsday for a leap year with my method, it is therefore simple to find Doomsday for any year.

Sun Mon Tue Wed Thu Fri Sat

---- ---- ---- 1900 1901 1902 1903

----19041905 1906 1907 ----1908

1909 1910 1911 ----19121913 1914

1915 ----19161917 1918 1919 ----19201921 1922 1923 ----19241925

1926 1927 ----19281929 1930 1931

----19321933 1934 1935 ----1936

1937 1938 1939 ----19401941 1942

1943 ----19441945 1946 1947 ----19481949 1950 1951 ----19521953

1954 1955 ----19561957 1958 1959

----19601961 1962 1963 ----1964

1965 1966 1967 ----19681969 1970

1971 ----19721973 1974 1975 ----19761977 1978 1979 ----19801981

1982 1983 ----19841985 1986 1987

----19881989 1990 1991 ----1992

1993 1994 1995 ----19961997 1998

1999 ----2000---- ---- ---- ----

Also, if you examine the table you will notice that the first year in a chain is always 6 more than another year with the same doomsday, the last year in a chain is 6 less than a year with the same doomsday, and the two years on either end are members of different chains. Understanding this relationship makes it very easy to come up with all the years in a century on which a certain date falls on the same day of the week.