Unexplained differences in the general classification

For each stage, I entered the stage results, the corresponding bonuses, and the penalties into a database. This allows me to calculate the expected final result. By comparing this with the final result, I determined which errors still remain in my database. I don't know exactly where the error is (otherwise I would correct it); it could be in the general classification, the stage results, bonuses, or penalties. Errors in secondary classifications are not indicated here, and I obviously can't find two errors that cancel each other out, and errors for riders who didn't finish the Tour are also impossible to identify this way. Nevertheless, I think this gives a pretty good indication of how complete my database is. The total error on December 19, 2018, is less than 22 hours, which is roughly the total time a rider spends 22% of a Tour. Since riders have finished the Tour approximately 15,000 times, the total error is roughly 0.0015%. In other words: 99.9985% is correct. (And I understand that this is a very misleading way to indicate accuracy. But at least I'm being honest that there's still some inaccuracy in my results.)

More information is available for some years than for others. The years for which little information is available are also those with the greatest uncertainty.

Roughly, it can be categorized as follows:

1903–1939
Archive L'Auto is online, so there is a lot of information. Not all penalties are mentioned, and there sometimes are minor inconsistencies, maybe because everything was calculated by hand.
1947–1987
Many complete daily standings can be found in El Mundo Deportivo, the Gazet van Antwerpen or Il Litterare. Exceptions are stages held on a Saturday or Sunday, because most newspapers were not published on Sundays or Mondays.
1988–1994
Almost always complete results in El Mundo Deportivo, exceptions are stages held on a Sunday.
1995–nu
Complete results on Cyclingnews or the Tour site.
Unexplained differences in the final general classification
YearTotalTop tenTop threeRiders with differencesStandings with doubtTime in doubt
19030:03.80 0:02.20 0:003 2 0:00.40
190430:00 30:00 0:001 1 0:00
19059369 Pts. 1678 Pts. 160 Pts.24 30 1:44
19060 Pts. 0 Pts. 0 Pts.0 0 0:00
19072 Pts. 0 Pts. 0 Pts.1 0 0:00
190822 Pts. 0 Pts. 0 Pts.3 0 0:00
19090 Pts. 0 Pts. 0 Pts.0 0 0:00
19100 Pts. 0 Pts. 0 Pts.0 11 0:00
19110 Pts. 0 Pts. 0 Pts.0 0 0:00
19122 Pts. 0 Pts. 0 Pts.2 0 0:00
19135:00 5:00 0:001 0 0:00
19142:04 0:00 0:002 1 0:00
19190:00 0:00 0:000 0 0:00
19200:00 0:00 0:000 0 0:00
19210:00 0:00 0:000 0 0:00
19220:00 0:00 0:000 0 0:00
192321:10 0:00 0:004 0 0:00
19240:30 0:00 0:003 0 0:00
19250:30 0:00 0:001 1 0:30
192616:29 0:00 0:007 4 0:00
192744:10 10:00 0:006 0 0:00
19281:44:40 1:01 0:0019 5 40:45
192915:48 10:00 0:009 1 0:00
19304:58 0:00 0:003 0 0:00
19312:00 0:00 0:001 0 0:00
19327:36 0:00 0:002 0 0:00
19330:40 0:34 0:003 0 0:00
19340:00 0:00 0:000 0 0:00
19352:52 0:00 0:004 5 2:20
19364:08 0:00 0:001 6 4:08
19373:05 1:00 0:004 26 0:06
19380:26 0:00 0:002 2 0:12
19392:48 0:00 0:007 13 2:07
19471:41 0:40 0:005 1 0:00
19480:30 0:00 0:001 0 0:00
19491:25 0:00 0:003 0 0:00
19503:53 0:00 0:009 0 0:00
19514:10 0:00 0:005 0 0:00
19522:19 0:00 0:007 1 0:00
19530:00 0:00 0:000 0 0:00
19541:00 0:00 0:002 4 0:00
19551:03 0:00 0:004 1 0:00
19563:00 0:00 0:005 4 0:00
19571:00 0:00 0:002 6 1:00
19580:30 0:00 0:001 0 0:00
19592:30 0:00 0:005 0 0:00
19603:47 0:00 0:005 0 0:00
19610:00 0:00 0:000 0 0:00
19626:57 0:00 0:004 1 0:01
19630:00 0:00 0:000 0 0:00
19640:00 0:00 0:000 0 0:00
19650:30 0:00 0:001 0 0:00
19660:00 0:00 0:000 0 0:00
19672:00 0:00 0:002 2 3:00
19680:00 0:00 0:000 7 0:00
19696:57 0:01 0:009 3 1:02
197046:31 0:00 0:0026 33 14:20
197121:30 0:00 0:0010 8 10:05
19728:52 0:00 0:0018 16 3:56
197315:30 0:00 0:0012 15 22:10
197432:31 0:06 0:0033 28 27:34
197522:01 0:00 0:0027 10 3:02
197644:29 0:00 0:0019 8 2:36
197712:25 0:10 0:0017 12 6:41
19784:35 0:00 0:0010 10 4:47
19791:10 0:00 0:004 0 0:00
198016:38 0:00 0:0014 11 4:59
198128:05 0:08 0:0041 7 2:00
19821:53:53 0:20 0:0087 8 1:58
198310:20 0:00 0:002 0 0:00
198410:49 0:00 0:007 8 1:01
19858:46 0:00 0:0024 10 2:20
19861:13 0:00 0:003 20 1:03
19870:06 0:00 0:002 4 0:20
198821:09 0:00 0:0039 41 20:27
19890:00 0:00 0:000 0 0:00
19904:32 0:00 0:007 19 7:59
19910:00 0:00 0:000 4 1:43
19920:00 0:00 0:000 0 0:00
19930:00 0:00 0:000 0 0:00
19940:00 0:00 0:000 0 0:00
19950:00 0:00 0:000 4 0:00
19961:59 0:00 0:005 0 0:00
19970:00 0:00 0:000 0 0:00
19980:00 0:00 0:000 33 0:00
19990:00 0:00 0:000 0 0:00
20000:00 0:00 0:000 0 0:00
20010:00 0:00 0:000 0 0:00
20020:00 0:00 0:000 0 0:00
20030:00 0:00 0:000 0 0:00
20040:00 0:00 0:000 0 0:00
20050:00 0:00 0:000 0 0:00
20060:00 0:00 0:000 0 0:00
20070:00 0:00 0:000 0 0:00
20080:00 0:00 0:000 0 0:00
20090:00 0:00 0:000 0 0:00
20100:00 0:00 0:000 0 0:00
20110:00 0:00 0:000 0 0:00
20120:00 0:00 0:000 0 0:00
20130:00 0:00 0:000 0 0:00
20140:00 0:00 0:000 0 0:00
20150:00 0:00 0:000 0 0:00
20160:00 0:00 0:000 0 0:00
20170:00 0:00 0:000 0 0:00
20180:00 0:00 0:000 0 0:00
20190:00 0:00 0:000 0 0:00
20200:00 0:00 0:000 0 0:00
20210:00 0:00 0:000 0 0:00
20220:01 0:00 0:001 0 0:00
20230:00 0:00 0:000 0 0:00
20240:00 0:00 0:000 0 0:00
20250:00 0:00 0:000 0 0:00

Total riders with differences: 591

Total difference in final classification: 12:19:14.79 and 9395 points.

Total difference in top ten: 59:02.20 and 1678 points.

Total difference in top three: 0:00 and 160 points.

I have doubts about 447 places.

Based on the information that I have now, I can calculate which stage results would give me the most information. If there were two stages on the same day, it is rare to find a full general classification after the first one. Also stages on a Saturday have less info because most newspapers don't publish on Sunday. The top twenty of stages with missing info is:

In the overview above, a mistake of 1 second counts the same as a mistake of 1 hour. This is solved with a different way of counting:

The corresponding total uncertainty 3:15:56.40. This is slightly different than the total mistake mentioned before, because definitions are slightly differen.

And finally, there is a list where a rider suddenly gets more or less time added than what follows from the stage results. This may indicate a time penalty. The stages where this happened most are:

And expressed in time:

The highest ranked rider for which not everything is correct is Louis Trousselier, who finished in rank 1 in 1905.