November 26, 1989

Pop Memories 1890-1954: The History of American Popular Music.  By Joel Whitburn.  Menomonee Falls, WI: Record Research Inc., 1986.  657 pp.

Reviewed by Tim Brooks

This is a dangerous book.

We’re all familiar with cases of inaccurate, sometimes wildly inaccurate, information that got into print under seemingly reliable auspices, then was repeated by careless authors until it became, in most people’s minds, fact.  It may take generations for the truth to catch up. Sometimes it never does.  Edison’s false date for the invention of the phonograph (August 12, 1877) comes to mind, as do many other bogus historical “facts.”  Here we have a whole book full of misleading information, presented in such a factual, almost statistical manner that it is bound to be quoted.  This has already begun to happen.  A presentation at the 1987 ARSC Conference was based in part on this book¹; a 1988 article in an academic journal drew significant conclusions from one of its more patently erroneous listings²; and it continued to be cited as a source in 1989³.

Pop Memories claims to be an index of the best selling record charts from 1890 to 1954.  It documents with great precision the best sellers by each artist during that 65 year period, showing each artist’s “chart” titles, the date each record reached the charts, highest position attained, and weeks on the chart.  The author, Joel Whitburn, is well known for his excellent series of indexes of modern Billboard record charts.  These are highly regarded, and deservedly so; they are models of precision, exhibiting the care of an indexer who double‑checks everything.  (Whitburn reportedly obtained original copies of every one of the 10,000‑plus singles that have placed on Billboard’s rock‑era charts, simply to verify label copy.)  I have been buying Whitburn’s Billboard indexes since they first came out, and would recommend them to anyone interested in the field.  They are carefully crafted, and get better with each edition.

Pop Memories, however, is another matter.  The problem, of course, is that there weren’t any best seller charts during most of the first 50 years of commercial recording.  Whitburn simply made them up.  (He implies, but never actually says this, in his misleading and hype‑filled introduction.)  Unfortunately he didn’t do it very well.

Although Pop Memories is being widely advertised and sold, apparently only one analytic review has ever appeared‑‑by this writer in 1987⁴.  It proved to be rather negative, which brought an angry and emotional response from the author.  Out of that exchange came a great deal of information about the book’s sources and methodology that had not been revealed to the book’s buyers.  Since the $50 volume continues to be actively promoted and, unfortunately, quoted (in fact a second edition is planned) it seems appropriate to revisit it here, with new information you won’t find in its introduction.

Misrepresented Sources

First a look at sources.  Many impressive‑looking primary sources are cited in the introduction (pp. 7‑10), but on close examination they do not contain the chart information the author says they do.  The Phonogram (1891‑1893), the industry’s first trade paper, is said to have provided “invaluable information on records and artists of the early 1890s.”  In fact this was a monthly that dealt almost exclusively with business matters and made only rare and vague references to the musical cylinders being produced then.  There are no record listings of any kind, no reports on individual record sales, and nothing that in one’s wildest imagination could be used to document ranked sales charts.

The Phonoscope (1896‑1900) is said to have “printed monthly lists of top popular recordings (although not in precise rank or chart form).”  This is simply not true.  The Phonoscope’s list was clearly titled “list of new records … sent us by the leading talking machine companies” and had nothing to do with popularity.  The author has since said he assumed a record was popular when its title was repeated from month to month; however The Phonoscope either repeated the entire list, or substantial contiguous chunks of it, not individual titles.  The publication was apparently repeating a company’s whole supplement when no new one was received.

“Sheet music sales” is cited as a source, but where this data was obtained is never stated.  There are no reliable sheet music sales figures for this period, and nothing was reported in any publication of the day on a regular basis.  The Phonoscope had a long column called “The Latest Popular Songs,” but this was compiled from lists supplied by the publishers themselves, hardly an impartial source.  This list was not ranked, and phonograph‑specific material (the monologues, minstrel routines, descriptives and similar material that sold so well) was excluded.  One suspects it was mostly a “plug list.”

“ASCAP and other lists of top period songs” is given as another source for the “charts” of the 1890s.  ASCAP was not founded until 1914 and has never published any sales data or rankings from this period.  Whitburn has subsequently said he used ASCAP’s “Hit Songs” brochures published half a century later, which give generalized lists of “big hits” of the 1890s.  ASCAP’s choices, of course, are totally undocumented.  Recent research suggests that they were based more on Tin Pan Alley legend and lore than on anything concrete.  Here are a couple of notable examples of songs that found their way into ASCAP’s “old favorites” brochures, thence into many books and finally on to Whitburn’s statistical charts.  “After the Ball” (1892) is shown in Pop Memories as one of the biggest hits of the decade.  The author, Charles K. Harris, spent the remainder of his long life loudly proclaiming what an enormous seller his song had been and how with it he had virtually invented Tin Pan Alley.  Even his autobiography is titled After the Ball.  Eventually others began to repeat his colorful stories.  But (1) examination of contemporary references in musical and theater journals indicate that while the song was indeed popular in late 1892 and early 1893, it quickly faded and was equaled or exceeded by scores of later, less‑remembered tunes; and (2) it had virtually no impact on records.   Another title that seems to have benefited from subsequent legend‑building is “Daisy Bell” (“On a Bicycle Built For Two”).  There is little contemporary evidence that it was anything more than one of many passing popular tunes.

Clinging to legend and lore, Whitburn imagines specific recorded versions of both these songs as “number one on the charts” for months on end and places both among the top ten best selling records of the entire decade.  He has since said that the songs must have been huge hits because they are mentioned in latter‑day books.  This is exactly how false information spreads.  An author should look to original sources, not secondary ones, to find out what actually happened.

Conversely, it should be noted, contemporary evidence suggests that “Ta‑Ra‑Ra‑Boom‑Der‑E” was a much bigger seller than either of the forgoing titles, at least on record.  While it is listed in Pop Memories it does not place as high as the forgoing‑‑apparently because its author did not tell colorful stories about it and it therefore gets less space in modern anecdotal histories.

Recent questions raised about Whitburn’s 1890s “charts” have resulted in the disclosure of a significant misrepresentation in the book.  According to later statements by the author, what the book consistently calls best selling “records” of the 1890s actually reflects best selling songs (also of questionable validity, as we have seen) and the heading “weeks on chart” really means “months.”  Why?  The author now says “We had to translate the reality of the 1890s record industry into the context of the 20th century record industry.  If, for example, we had listed (a record) as having been #1 for 50 weeks, it would so totally skew chart history as to make it incomprehensible in modern terms” (author’s emphasis)⁵.  In other words, people wouldn’t understand what really happened, so we changed it to look more like today.  This is an absolutely appalling way to approach historical research, especially when the author doesn’t tell the reader he’s doing it!  The historian’s job is to reconstruct what actually happened, not change it to look like last week’s Billboard charts.

More on sources.  The articles of Jim Walsh are given as “the single most important source for the (1900‑1920s) charts.”  Jim is one of the legends of pioneer artist research, and a longtime friend of this writer, but he wrote biographical sketches not technical articles about record sales.  There are mentions of best sellers within the biographies, of course, but these are almost always anecdotal and hardly the sort of thing from which weekly (or is it monthly?) ranked charts could be constructed.  When Jim did give specific figures he was prone to the same mistakes as anyone else.  He once repeated in print the old story that Vernon Dalhart’s Victor recording of “The Prisoner’s Song” had sold six to seven million copies.  I asked Jim where he got that, and he replied rather sheepishly that he probably shouldn’t have said it; he knew little about “such things” and the figure, which came from a Victor promotional flyer printed in the 1940s, seemed highly improbable.  In fact figures compiled by Victor and reproduced in several modern sources indicate that the record’s sales were a fraction of this amount, probably about one million copies⁶.  Whitburn, however, repeats the seven million figure as fact and ranks the record right after “White Christmas” and “Rudolph the Red‑Nosed Reindeer” as the third‑biggest seller in history.  No sources are given.

Talking Machine World is cited as containing “monthly lists of nearly all popular record releases” starting in 1905.  These had nothing to do with popularity.  Then, the author says (page 7), “from 1914‑21 the major record companies provided monthly lists of their best sellers to Talking Machine World.”  I searched long and hard for these lists; after years of using TMW as a source I had never seen anything like that.  Apparently Whitburn was referring to a monthly top six for Chicago only, which appeared in the local news round‑up for that city.  Needless to say this list was very limited in scope. TMW had no national list and no references to sales in its editorial columns.  Whitburn’s characterization of this source is misleading in the extreme.

In fairness to the author, information on individual record sales did become more available in the 1920s although there is still hardly enough to justify a weekly chart. (Whitburn hasn’t said whether his heading “weeks” means weeks or months in the 1920s.)  The introduction says “Billboard and Variety provided abundant information on the hit records and songs of the 1920s.”  The author apparently relied on occasional mentions of sales in news columns.

In 1929 Variety began printing a monthly ranking of best sellers by label, based on information supplied by the labels themselves.  While subject to the biases of what each label wanted printed, no ranking of labels against each other, and the omission of all minor labels, this constitutes the best regularly published information to date on individual sellers.

Additional rankings came and went during the early 1930s, with radio’s Your Hit Parade inaugurating the idea of a weekly national tune parade in 1935.  All of these are subject to biases of one kind or another (it has never been revealed just how Your Hit Parade’s sponsor‑controlled ranking was computed) but they do give a pretty good idea of month‑by‑month and even week‑by‑week song popularity in the 1930s.  Translating this into individual record popularity requires some large leaps, but it can be done.

From 1940‑on Whitburn has been able to use Billboard’s national charts, which were inaugurated in that year.  (These were previously indexed in his valuable book Top Pop Records, 1940‑1955.)  One can argue about where Billboard got its information, but at least from this point forward the book is based on a clearly defined and presumably impartial primary source; the author did not have to “make up” the charts.

How Accurate Are the “Made‑Up” Charts?

Since Whitburn obviously didn’t get his early information where he said he did, on what were these charts based?  They appear to be guesswork, reflecting modern biases and a misunderstanding of how the record industry operated in its formative years.  This is increasingly true the further one goes back.  Thus the representation of the 1890s is ludicrous, that of the 1920s merely erroneous in particular listings.  Whether inadvertently or intentionally the author, who is most familiar with the rock era, has forced earlier eras to conform to the chart patterns with which he is familiar, and has favored records and artists that appeal to collectors today.

Cylinders of the 1890s are shown moving up and down the “charts” with alarming rapidity, much as rock records do today.  The first number one is “Semper Fidelis” by the U.S. Marine Band for six weeks, followed by “Washington Post” for six and “The Thunderer” for four.  None of these ever returned to the “charts” to sell more because today rock records don’t do that.

This is not how cylinders were distributed in those faraway days.  There is ample evidence that a few best sellers sold steadily (mostly to coin‑machine operators) for very long periods of time.  The catalogs of Columbia and other companies bear this out, and there are even a few sales figures available.  In December 1892 the Phonogram reported that during the prior two years the “Pat Brady” monologues by Dan Kelly for the Ohio Phonograph Co. had sold an amazing 5,000 cylinders.  Whitburn garbles this as “Pat Kelly” and has him on and off the charts in about two months in 1891.  Several sources from 1895‑1896 indicate that the two comic specialties of George W. Johnson, the first black recording artist, had sold a phenomenal 25,000‑30,000 copies during the first half of the decade and were still major hits.  Whitburn pushes Johnson’s two titles on and off the “charts” during the single year 1891.  This is nonsense.

Fagan and Moran’s invaluable Encyclopedic Discography of Victor Recordings, Vol. 1, shows the press runs of many early Victors from 1900‑1903.  Combining this information with the number of issued takes for each title (a predictable 500‑1500 copies could be pressed from a single master then) one can derive a picture of what was selling.  Of the six records with the largest press runs in EDVR only two show up on Pop Memories’ “charts,” neither as a very big hit.  The largest total I could find in EDVR was 10,124 copies of a song called “Truscalina Brown” by Silas Leachman (6 takes).  This is not listed at all in PM.  The two titles by Leachman that are listed had press runs of about one‑tenth that total.

A certifiable best seller of this period is “The Holy City.”  The book shows Harry Macdonough’s very familiar version reaching “number two for one week” in late 1900, then disappearing.  When this was pointed out the author dismissed the title as an “old time evergreen” which probably sold “slowly but steadily for many months or even years.”  He provides no basis for this statement, and Victor’s files paint a very different picture.  Macdonough was called back repeatedly to record new takes of this number, in June 1900, November 1900 (twice), February 1902, June 1902, September 1902, January 1903, February 1903, April 1903, July 1903, November 1903, May 1904, June 1904, August 1906 and December 1908.  In addition, during 1900‑1903 alone Victor had nine of its other artists record and re‑record the title.  Does this sound like a “slow, steady seller”?  Number two for one week?  “The Holy City” was probably Victor’s number one best selling record for three or four years running.

Production data is available from the Columbia files for the period after 1915.  Of that label’s 14 top selling records from 1915‑1920 only five turn up on Whitburn’s “charts,” three of them briefly and at lower chart positions.  In other words, Whitburn got only two out of the 14 “right” (as hits), putting instead scores of lesser sellers at the top of his rankings.  The author, incidentally, has since claimed his charts to be “more than 90% accurate” for the period prior to 1930⁷.  The above sample suggests that 14% (two out of 14) might be closer to the truth.

There are similarly unreasonable listings in the 1920s.  A recent scholarly article based part of its analysis on Pop Memories’ listing of the socially relevant title “The Farm Relief Song” as a top ten hit in 1929⁸.  Columbia’s files show that this record shipped about one‑sixth the total of first‑line pop releases that fall.  Twenty‑three different pop tunes released by the label in the same month all outsold it (although only four appear on Pop Memories’ “charts”).  Collectible artists frequent Whitburn’s best sellers chart during the 1920s: Mamie Smith on the poorly distributed Okeh label (eight top hits), Fletcher Henderson’s jazz orchestra (nine), Fiddlin’ John Carson (six), Red Nichols (six), Louis Armstrong’s Okeh electrics (eleven), King Oliver (five), even Bix Beiderbecke (two hits, including the super‑rare “In a Mist”).

The bizarre omissions and inclusions follow a certain pattern, which gives us a clue as to what went wrong with these charts.  Columbia’s top sellers during 1915 and 1916 were almost all Hawaiian numbers.  These are not considered very interesting or collectible today and they are severely underrepresented in Pop Memories.  On the other hand Al Jolson is very collectible, and four of his relatively rare Victors from 1912‑1913 are shown on the charts, three reaching number one.  Classical artists are also favored.  Victor’s Geraldine Farrar is shown with ten chart‑toppers between 1907 and 1916, and Columbia’s Louis Graveure with three (all top ten hits) in the late teens.  These prestigious artists were heavily promoted by their labels, but this does not mean their discs outsold popular and dance numbers.  The Columbia files indicate that most of Graveure’s releases sold no more than 3,000‑4,000 copies; at the same time (1916‑1920) Louise and Ferera’s “Drowsy Waters” shipped 322,000 copies, but Whitburn does not list it at all on his charts.

There are many more examples of this pattern‑‑interesting, collectible records are given high chart positions, uninteresting titles are downplayed.  The use of anecdotal sources is also evident.  For example, Art Landry’s 1923 Gennett recording of “Dreamy Melody” is shown as reaching number one.  Landry long regaled listeners with tall tales about how his first recording sold “one and a half million copies,” which was probably more than the entire annual production of the tiny Gennett label.  The claim was repeated by irresponsible authors and finally turns up here, “at the top of the charts”!  Try to find a copy of this “hit” today.

What Can Be Done?

Is it possible to compile a reasonably accurate book on record popularity during the first fifty years of the industry?  I believe it is, but it wouldn’t look like Pop Memories.  There is simply not enough data to support such precise rankings, and to suggest otherwise is misleading.  One possible approach might be to compile alphabetical lists of the most popular records by year from, say, 1890‑1910, then by month after that (with records perhaps grouped by general popularity level in later years).  Not until the 1930s would ranked weekly charts seem to be justified.  This is not as neat or seemingly precise as the Pop Memories approach, but it is truer to the actual source data.  It is, in a sense, how the published charts themselves evolved.

As for sources, primary reliance could be placed on three major indicators.  Most important is internal company pressing, shipment and sales data, where it exists (Victor, Columbia, etc.)  Whitburn was apparently not aware of this information.  Second, a census of which records show up most frequently today in large, general collections.  If you find ten copies of “The Holy City” by the Haydn Quartet on Victor with a 1905 take and label type for every one of Billy Murray’s “Give My Regards To Broadway” on Columbia, it is safe to assume that the former outsold the latter in that year.  (Whitburn shows the opposite ratio for these two titles; I’m not sure which is right.)  Such a census, done right, would be time‑consuming and difficult, but not impossible, and would represent a real breakthrough in knowledge.

Third, for the 1890‑1920 period one could index company catalogs to see which selections were listed in catalog after catalog, had multiple versions, were included in medleys, mentioned in parodies, and so forth‑‑all indicators of popularity.  Conversely, one would see which titles were quickly deleted.  Such data would be especially valuable in the pre‑1920 era, when long‑term sales were characteristic of the “hits.”

Also, a compiler should look at many of the sources Pop Memories consulted, particularly items in the contemporary press which offered specific information‑‑always being wary, of course, of the reliability of the source and the dangers of fragmentary information.  There is a good deal more data out there than is sometimes supposed, even though it is not of the precision Whitburn indicates.

Does This Book Have Any Value?

 If Pop Memories contains so many misleading claims and demonstrable errors, and is so hopelessly skewed toward the “rock perspective” of record history, is it of any value at all?  Yes.  If you strip away all the nonsense about chart positions and “number one with a bullet” it is worthwhile as a selective listing of some of the generally more popular recordings and artists of the 78 rpm era.  Any book that gives us six pages of Billy Murray records, year by year, with labels and record numbers and a short bio making clear his importance in the recording world, can’t be all bad.  In fact, most of the important artists are here, including many who are seldom recognized elsewhere.  A top‑artists‑of‑all‑time ranking, based on number and magnitude of chart hits, is speculative and ought to be labeled as such, but it is not entirely unreasonable.  It’s simply Whitburn’s opinion.