Date Drift

By changing to a new calendar Pope Gregory began a process that has obscured when events in history actually happened. This is known as "drift" error.


The best way to explain Drift Error is by example...

George Washington’s Birth

The story most often used to show the problems of reporting errors is the date of George Washington’s birth. George was born February 11, 1731 on the old style Julian calendar. After the calendar revision in 1752, George was now born on February 22, 1732 on the new style Gregorian calendar. His birthday had not really changed, but the label given it had. At the time of transition the dates were indicated with "OS" for old style and "NS" for new style, but as we have forgotten this transition we often leave off the calendar used to report the date.

Leaving off the calendar on any historical date is the root of drift error when dates are reported.

Trust no date from this or any previous period unless the calendar used to report the date is also clearly identified.

The following shows an Auditable Date Report on George Washington’s actual birthday. Notice that both the OS and NS calendars are showing their respective labels for what in history was 1 actual day, the day George was born. Notice that the report shows the day of the week in this case a Friday, and notice also that there are 4 other calendar dates assigned to this same day.

George Washington’s Birthday
Fri12722-09-19 AA22 Feb 1732 NS11 Feb 2485 AUC4640869 AAN
3222-09-19 FE11 Feb 1731 OS11 Feb 1732 AD2353712 JDN

Because Gregory’s calendar conversion changed the date for New Years Day, the year number changes for dates between January 1 and March 24.

Any date that falls in this period runs the risk of not being off by just 11 days, or so. Dates in that part of the year risk being off by whole years or more.

Clues that Drift Error is occurring

Drift Error is occurring is when a historical account indicates the "close of a year" at the end of December yet the subject year is before Gregory’s introduction of the calendar in 1585, but after the adoption of the Christian year conventions. It was not lived that way and the report is based on drift error.

Drift Error is occurring when the day of the week is incorrect. Curiously, the day of week is like a computer check-sum. Dates that exhibit drift error loose their correct day. If the day does not agree with the results of a query such as that shown here drift error is occurring.

Continued Example

To see how Drift Error propagates, consider if you were given the date for George’s birthday, February 11, 1731, and were not told the calendar, and you assumed that it was the current, modern, NS calendar.

Of course it may not be you making the mistake. Anyone along the chain of documents reporting the date to you can make this mistake. This is typically when Drift Error mistakes are made.

The following is an example George’s birthday, but with 1 round of drift error. The error, in this case was to assume the wrong calendar.

George Washington’s Birthday, IN ERROR
Sun12721-10-03 AA11 Feb 1731 NS31 Jan 2484 AUC4640493 AAN
3221-10-03 FE31 Jan 1730 OS31 Jan 1731 AD2353336 JDN

The difference between these 2 dates is 4640869 - 4640493 = 376 Days.

This is the total error in this example based on not correctly reporting the calendar used to tracking the date. The error, in this case is over 1 year.

Note that Sunday is not the same as Friday, the actual day of the week when George was born. This is one of the clues that indicate drift error is happening. It can be spotted, though, only when you know the day of the week when something happened.

Additional Drift Error

This case with George’s birthday is insightful because there is lots of documentation on his birth, so we have a golden reference from which to measure the error.

But, what if, there was only one document that gave his birthday, and it was edited by someone who did not know with certainty when the event happened because there was no calendar notation.

Imagine a copyist made the conversion again. Say, they recognized that the February 11, 1731 birthday was NS and that it must have happened on the OS calendar, since that is what was used at his birth, so they convert this to the OS calendar and report that George was born on January 31, 1730.

Of course this could also be miss-interpreted as a NS date and the following would be the assumed date for George’s birth.

George Washington’s Birthday, IN ERROR AGAIN
Tue12720-09-17 AA31 Jan 1730 NS20 Jan 2483 AUC4640117 AAN
3220-09-17 FE20 Jan 1729 OS20 Jan 1730 AD2352960 JDN

The difference between the original, known date for George Washington’s birthday and the day reported here is 4640869 - 4640117 = 752 Days.

Notice that the error is going up based on the generations in the documents.

For all events in the era before the use of the modern calendar the error bounds on any date is related to the number of generations in the documents back to the original event.

The number of generations in most documents is unknown.

So the error is unknown and theoretically unbounded.

In other words: because there has been a calendar change we no longer know with certainty when anything happened.

Pre-Jesus Roman Calendars

In the year 46 BC the Roman world corrected calendars in a dramatic fashion. The calendar conversion problems were significantly greater than touched on in this report.

The risk of Drift Error is thus considerably higher for all BC dates. No dates from the era before 46 BC should be considered trustworthy.