[Dixielandjazz] ODJB in 1917

Anton Crouch a.crouch@unsw.edu.au
Thu, 15 Aug 2002 17:43:17 +1000

Hello Steve and Bill

No, please don't agree to disagree. Let's look at the evidence and come to
a decision.

There are 4 reasonable hypotheses about what happened between the ODJB and
Columbia in 1917, viz:

1. The ODJB visited the Columbia NYC studios once only, on or about 31
January 1917, and recorded 7 takes (4 of "Darktown strutters ball" and 3 of

2. The ODJB visited the Columbia NYC studios twice in 1917, once on or
about 31 January 1917 and again on or about 31 May 1917. Recordings were
made on both occasions but the allocation of dates to take numbers is not

3. As for 2 but no recordings were made on the January 1917 visit, ie the 7
takes of the two titles were made on or about 31 May 1917.

4. The ODJB visited the Columbia NYC studios once only, on or about 31 May
1917, and recorded the 7 takes of the two titles.
How do these hypotheses stand up in the face of the evidence?

1 is highly improbable. The 77000 matrix numbers were not used until April
1917. Is it possible that earlier matrix numbers were removed from the
waxes and later numbers added (when Columbia decided to issue the
recordings)? If so, why is there no reference to such a procedure in the
files? Also, what were the earlier numbers? There is no gap in the 47000
sequence for the January-February 1917 period and nothing by the ODJB has
turned up in the 60000 to 64000 test and personal record series. Believing
hypothesis 1 involves believing either that no matrix numbers were
allocated at the time of recording or that two matrix numbers were used
from a series that didn't exist at the time of recording.

2 is also highly improbable, for the same reasons that apply to 1. 2 has
the additional problem of being in conflict with Columbia's practice of
"carrying over" matrix numbers when recordings were made on subsequent
dates. If the ODJB had recorded various takes of the same titles on
different dates, the originally allocated matrix numbers (in the 47000
series for a January 1917 date) would have been retained for the later date
with the take numbers continuing sequentially from the first date.

3 fits all the available evidence.

4 is, discographically, the simplest hypothesis but it fails when the Eddie
Edwards evidence is considered.

So, let's not "agree to disagree". Let's accept hypothesis 3 until such
time as either it is shown to be wrong or a more parsimonious one is proposed.

All the best

PS: The matter of the musical worth of the 1917 Columbias is subjective and
not evidence in this discussion. I actually prefer the first Columbias to
the first Victors - as Bill says, more relaxed.   :-)