Charles Dow and the Birth of the Stock Index, 1884–1896

On 3 July 1884 a two-page sheet called the Customer's Afternoon Letter, written out by hand in a Wall Street basement and carried to subscribers by messenger boys, printed a number that had not existed before. It was 69.93. It was the sum of the closing prices of eleven stocks — nine railroads, a steamship line and a telegraph company — divided by eleven.

Nothing about the arithmetic was hard. Any clerk with a pencil could have produced it, and clerks had been producing something like it privately for years. What had not been done was the decision to publish the figure every trading day, for the same list, in the same form, so that a subscriber could set today's number against yesterday's and answer a question that until then had no answer: did the market go up?

A Basement Next to the Exchange

Charles Henry Dow was born on 6 November 1851 on a farm in Sterling, Connecticut, and lost his father at six. He had no college education and no training in finance. What he had was fifteen years of newspaper work — the Springfield Daily Republican under Samuel Bowles, then the Providence Star, the Providence Evening Press, and from 1877 the Providence Journal, where he wrote a long historical series on the steamship business of Narragansett Bay.

In 1879 the Journal sent him west with a party of bankers, lawyers and mining promoters to Leadville, Colorado, then the loudest silver camp in America. He filed nine letters from the mountains. They are worth reading because of what they are not: there is almost no colour in them, no prospector folklore, very little about the saloons. Dow wrote about assay values, capitalisation, the terms on which shares were being sold to eastern investors, and the gap between the two. He had found the subject he would spend the rest of his life on, which was the distance between what a security cost and what it was worth.

He moved to New York in 1880 and joined the Kiernan News Agency, a financial wire service that served bankers in the district by messenger. Edward Davis Jones, a Brown University dropout he had known in Providence, came with him, and a third man, Charles Milford Bergstresser of Lafayette College, joined them. Bergstresser put up much of the capital and was by most accounts the best reporter of the three, able to get bank presidents to talk. His name was left off the door because Dow, Jones and Bergstresser would not fit on a letterhead (Wendt, 1982).

Dow, Jones & Company opened in November 1882 at 15 Wall Street, in a basement room behind a soda-water and candy shop, a few doors from the entrance to the New York Stock Exchange founded ninety years earlier by the twenty-four brokers of the Buttonwood Agreement. The firm's product was news, delivered fast. Reporters carried scraps of paper back to the basement, where copy was written and flimsies were run off on a hand-cranked press, and boys ran the bulletins to brokerage houses within minutes. By 1884 the accumulated bulletins were being bound each afternoon into the Customer's Afternoon Letter, which reached roughly a thousand subscribers.

Eleven Names, Divided by Eleven

The first average was a railroad average because in 1884 the stock market was railroads. Industrial companies were mostly partnerships or closely held; their paper traded thinly, on the curb or by private treaty, and carried a reputation somewhere between speculative and disreputable. Rails were the listed market, and rail bonds were what trust departments and insurance companies actually owned, as the collapses of Jay Cooke's railroad financing in 1873 had already demonstrated to a generation of investors.

Dow's eleven were Chicago & North Western; Chicago, Milwaukee & St. Paul; Delaware, Lackawanna & Western; Lake Shore; Louisville & Nashville; Missouri Pacific; New York Central; Northern Pacific preferred; Union Pacific; the Pacific Mail Steamship Company; and Western Union. The last two were not railroads, which mattered later: the average was never conceived as a pure sector measure but as a sample of the securities that carried the market's weight.

The method was a simple arithmetic mean of closing prices. Add eleven numbers, divide by eleven, print the result. Dow chose it because it was transparent and because a subscriber could check it, and both of those properties turned out to matter more than statistical elegance. A weighted index would have required capitalisation figures that were not reliably published; an unweighted price average required only the tape.

The Wall Street Journal

On 8 July 1889 the Customer's Afternoon Letter became a newspaper. The Wall Street Journal appeared in four pages at two cents a copy, with Dow as editor and a staff of about fifty. Its opening statement of purpose was a promise of restraint: the paper would give full and fair information on the fluctuations in prices of stocks, bonds and some classes of commodities, and it would aim steadily at being a paper of news and not a paper of opinions.

That sentence was a commercial position as much as an editorial one. Financial journalism in the 1880s was largely bought. Tip sheets took payments from pool operators, and a favourable paragraph about a stock could be had for cash. A paper that printed the closing prices, the average, and nothing it had been paid to print was selling something scarce. Circulation reached roughly seven thousand by the end of Dow's tenure.

Dow also, briefly, sat inside the market he covered. He bought a seat on the New York Stock Exchange in December 1885 in order to become a partner in the brokerage of Goodbody, Glyn & Dow, and sold it in April 1891. Those five years gave him the floor's vocabulary and an understanding of how orders actually filled, which shows in the precision of the editorials he wrote a decade later.

Twelve Industrials

By the mid-1890s the composition of the listed market had changed. The trust movement had turned family manufacturers into corporations with public paper — sugar, tobacco, lead, rubber, distilling, electrical equipment — and their shares were trading in volume that no longer looked like the curb. A railroad average was measuring a shrinking share of what the market did.

On 26 May 1896 the Wall Street Journal published the Dow Jones Industrial Average for the first time. It contained twelve stocks and stood at 40.94.

Five months later, on 26 October 1896, the railroad list was formally separated into a twenty-stock Dow Jones Railroad Average, a measure that survived under that name until it was renamed the Transportation Average in 1970. From that autumn the paper was publishing two averages, and the pairing was deliberate: Dow's later editorials treat agreement between the two as the only reliable confirmation that a market move is real.

The industrial average's first months were brutal. Cleveland's gold reserve crisis and the free-silver campaign of William Jennings Bryan drove the new index to 28.48 on 8 August 1896 — a level it has never revisited and never will, a low set ten weeks after the series began.

The Flaw Built Into the Arithmetic

A price-weighted average has a defect that was obvious from the start and has never been repaired. A stock trading at $200 moves the number four times as hard as a stock at $50, whatever the size of the two companies. A split changes a company's influence on the index without changing anything about the company.

Dow handled substitutions and splits by adjusting the figures by hand, which introduced discontinuities nobody could reconstruct afterwards. A systematic fix arrived only on 1 October 1928, when the list expanded from twenty to thirty names and the compilers replaced the divisor — until then simply the number of stocks — with a variable that is recalculated on every split and substitution so that the printed level does not jump. That divisor has since fallen below one, which means the average is now a multiple of the summed prices rather than a fraction of them.

Every serious index built afterwards rejected the weighting. Standard Statistics began capitalisation-weighted indices in 1923, and the 500-stock composite that became the benchmark for John Bogle's first index fund in 1976 weights each company by its market value, which is what an investor's portfolio actually does. Jeremy Siegel's long-run return series relies on capitalisation-weighted data for precisely this reason (Siegel, 1994).

The Dow survived anyway, for a reason that has nothing to do with statistics. It was first, it was published daily without interruption for over a century, and by the time better measures existed the number had become the thing the public meant by "the market." When the ticker stood at 230.07 on the afternoon of Black Tuesday in October 1929, or at 1,738.74 after the machines broke the market in October 1987, it was the Dow that carried the news.

Three Movements

Dow never wrote a book and never used the phrase "Dow Theory." What he left was roughly 255 unsigned editorials in the Wall Street Journal between 1899 and his death in 1902, written in a plain, cautious register that reads less like market commentary than like a careful man describing a machine.

The central passage appeared on 4 January 1902. "The market is always to be considered as having three movements, all going on at the same time," he wrote. "The first is the narrow movement from day to day. The second is the short swing, running from two weeks to a month or more; the third is the main movement, covering at least four years in its duration." The claim underneath it is that the daily noise is not the signal, and that the average exists to let a reader see past it.

He returned repeatedly to an image of the tide. In an editorial of 20 July 1901 he described a person on a beach who wants to know whether the tide is still coming in, setting a stick in the sand at the point the waves reach and moving it up until the waves stop reaching it. Watching a single wave tells you nothing; the stick records the direction.

Samuel Nelson gathered fifteen of the editorials into The ABC of Stock Speculation in 1903 and gave the ideas their name (Nelson, 1903). William Peter Hamilton, who succeeded Dow as editor of the Journal, spent twenty-seven years applying them in print and codified them in The Stock Market Barometer in 1922, adding the formulation the theory is usually remembered by — that the averages discount everything (Hamilton, 1922). Robert Rhea systematised Hamilton's rules further in 1932, at the exact bottom of the decline that had taken the industrial average from 381.17 to 41.22 (Rhea, 1932).

Whether any of it worked is one of the few questions in technical analysis to receive a serious empirical answer. Brown, Goetzmann and Kumar reconstructed Hamilton's twenty-seven years of editorial calls, fed them to a neural network to make the decision rule replicable, and tested the resulting strategy against buy-and-hold over the same period. Hamilton's timing produced higher risk-adjusted returns, mostly by being out of the market during declines (Brown et al., 1998). The authors were careful to note that this is a finding about one man over one sample, not a licence.

After the Basement

Dow sold Dow Jones & Company to Clarence Barron, the Boston correspondent who had been the firm's largest customer, in 1902 for about $130,000. He died on 4 December that year, aged fifty-one, at his home on Brooklyn's Pacific Street. The Journal's obituary ran on an inside page.

He had been dead five years when J. P. Morgan assembled the rescue described in the Panic of 1907, and the newspapers covering it quoted the industrial average as a matter of course — a fact that would have been impossible twenty-five years earlier, because there had been no number to quote. Richard Stillman's institutional history of the index counts the substitutions, the divisor changes and the compositional drift, and concludes that almost nothing of the 1896 list survives except the practice of publishing the number (Stillman, 1986).

That practice is the invention. Before July 1884, the market's condition was an opinion — something a broker asserted and a customer either believed or did not. After it, the market's condition was a figure that anyone could look up, argue with, chart, and check against the figure of the day before. Index funds, benchmarks, beta, the entire apparatus by which performance is measured against something rather than against nothing, all descend from a clerk's arithmetic on eleven railroad stocks, published because a subscriber paying for an afternoon letter deserved to be told what had happened.

General Electric, added to the industrial list on 26 May 1896 and removed for the last time on 26 June 2018, outlasted every other name Dow chose by more than sixty years — and the average that dropped it is still printed each afternoon, still the sum of prices divided by a number.