A side (mx. B-3049, take 1) recorded Philadelphia, PA (?), January 26, 1906.
A side (mx. B-6848, takes 1 or 2) recorded New York, NY, March 2, 1909.
B side (mx. B-2885, takes 1 or 2) recorded Philadelphia, PA (?), November 14, 1905.
Cutout date: January 1925.
(DAHR)
Both tracks had originally been released on single-sided Victors,
"Buffalo Rag" on Victor 4628.
"A Barnyard Serenade" on Victor 4562.
Thanks, fixbutte, for explaining the reason for this release date to everyone. It always pays to look at original contemporary sources like the Talking Machine World. This particular record is not mentioned in the Victor listings for December 1910 in The Talking Machine World, Nov. 15, 1910, page 54 because it was really just a double-sided reissue of tracks that had earlier been released as single-sided records. But one can see that the catalog numbers fits between the block of catalog numbers announced there, even though they apparently were released out of numerical order.
This record was extremely popular, being repressed from 1910 to 1925. Not surprisingly, "Buffalo Rag" is considered one of the best and most authentic ragtime recordings of one of the best pieces by pioneering African American composer Tom Turpin. The labels currently illustrated represent a Spring 1918 to Fall 1923 repress on Victor Batwing label without price notice, listing four patents (1903, 1904, 1905, and 1908), cf. M. W. Sherman, Collector's Guide to Victor Records, 2nd ed., page 84.
DAHR claims that the 1905 version of "Buffalo Rag" was still used for some issues of this 1910 record. I find this hard to believe, considering that Ossman had to re-record the tune as early as 1909, probably because the stampers of the original version had all worn out.
So I'd love to know what the take numbers on your copies are, vocalion red and Matthew Rose. You'll find them in the 9 o'clock position of your runouts.
mduval32323's quote is from the correction queue (I hope he doesn't mind). I just wanted to clarify why his correction had to be rejected, and I did it in the comments because it may be of general interest (another mod had already approved the correction).
@mduval32323
I've corrected the release back to November 1910 according to the release date given in DAHR, which you find impossible because "the entire Columbia 167xx catalog series was 1911 releases and this was the 79th one in that series", so it was more likely April 1911 or just 1911.
Anyway the Talking Machine World issue of November 1910 already shows non-consecutive Victor numbers up to 16833 (including 16789), so your premise is disproved,