The History of the Organization Chart

By Victor Hoang · Updated Oct 23, 2026

Picture the problem in 1855. Daniel McCallum told the Erie's president that a short road could get its superintendent's personal attention. A long road was a different matter. Its failures came from the lack of a system, not from its length, as Chandler quotes him.

The organization chart as a printed diagram dates to the 1850s. McCallum, general superintendent of the New York and Erie Railroad, had his divisions and reporting lines drawn on one sheet. The business historian Alfred Chandler calls it one of the earliest organization charts in an American business. Journalists often call it the first. The throughline: a chart appears when a business grows past what one person can see. It spreads when big firms copy a working template. And it keeps being tested against how work actually flows.

In McCallum's telling, the superintendent of a road fifty miles long could give it his personal attention, but a road five hundred miles long was a different matter. The Erie's divisions ran from Jersey City to Dunkirk on Lake Erie, and McCallum wrote that shorter lines offered no plan he could simply borrow. This article follows what he drew, who copied it, who taught people to draw it, and what the diagram leaves out. For the wider story, see the history of management.

Timeline at a Glance

The Chain of Hands (antiquity-1852)

Orders pass down tiers long before anyone draws them

Start with a desert camp and a tired judge. In the Book of Exodus, Moses sits from morning until evening settling disputes, and his father-in-law Jethro tells him, in the Bible Gateway NRSVUE text, "What you are doing is not good." His fix is to choose capable, trustworthy men and set them over groups of thousands, hundreds, fifties and tens. They decide the routine cases and bring only the hard ones upward. The passage has no firm date, and it is a story, not an administrative file. But it is a well-known early record of delegation: tiers of leaders, with an escalation path.

A second record is military. Polybius, writing in the second century BCE, sets out how the Roman legion was officered in Book 6 of his Histories. In his account a legion in normal times held 4,200 infantry and 300 horse. Ten centurions were chosen "according to merit" from each class of troops except the youngest, each company had two centurions, and the centurions then appointed an equal number of rearguard officers, the optiones. Ranks, counts and a way of picking leaders are all there. What is missing is a drawing.

These were precursors, not org charts. Neither text shows the structure as a picture, and neither deals with the thing that drove the 1850s change: managers who had to watch other managers across hundreds of miles. Before the railroads, Chandler writes in Strategy and Structure, even a railroad was rarely more than fifty or a hundred miles long, and its general superintendent could keep close contact with everyone who carried out the road's business.

What this means for teams today: Delegation by tiers is old, and so is the escalation path. If your team cannot say which decisions stop at which level, you have Jethro's problem without his solution.

The Erie Plan (1853-1857)

A railroad superintendent puts authority on one sheet

The scene is a general superintendent's office in New York. McCallum had come up as a bridge builder, and, by Michael Dearing's account, he was born in Scotland in 1815 and moved to the United States at seven. Chandler explains why the Erie needed him. After its completion in 1851 the road suffered high operating costs, and in spring 1853 the Erie Canal's short lines merged into the New York Central, making that route a stronger rival. That autumn the Erie's board asked for a reorganization that would give more precise accountability for expenses and a better appraisal of managers. McCallum was promoted to general superintendent to do it.

His answer had two parts. The first was a short list of principles. In his 1855 report he set out six: a proper division of responsibilities, enough power to carry them out, a way of knowing whether they were done, prompt reports of failures, daily reports and checks that would not embarrass senior officers, and a system that would let the general superintendent detect errors at once and name the person responsible. He added that subordinates should answer only to their immediate superiors, a one-boss rule that Fayol would name unity of command six decades later. The second part was information. Hourly reports travelled by telegraph, locating trains and explaining delays, and daily and monthly reports gave costs per locomotive, as Chandler's chapter on the railroads describes. A plan of organization showed who reported to whom.

That drawing is the document usually called the first modern organization chart. It was drafted by a civil engineer, George H. Henshaw, and Slate reports that it gave daily responsibility to the divisional superintendents, who reported to McCallum and the central office. A copy of the 1855 diagram sits in the Library of Congress. Visually it follows a tree, and Slate notes that from a distance it looks like a tree with fruit or blossoms. Chandler is more careful about priority than the headlines. He calls it "certainly one of the earliest organization charts in an American business enterprise." Slate calls it the first formal one. Both point to the same Erie chart, and the difference is only in the wording: "first" against "one of the earliest."

The chart spread quickly. Poor, editor of the American Railroad Journal, wrote in 1854 that the Erie's superintendent could tell at any hour where every car and engine was, and he later had the chart lithographed and sold copies for a dollar. A British railway expert described McCallum's work in a parliamentary report printed in 1857, and the Atlantic Monthly praised it in 1858.

The Erie itself could not hold on to the system. Chandler says that before the decade ended the road fell into the hands of financiers like Daniel Drew, who cared little about administration, and McCallum left. His methods were tested on the Pennsylvania instead. There, in December 1857, President J. Edgar Thomson signed an organization manual that borrowed many of McCallum's words and phrases but gave division superintendents the authority to order people in the other functional departments on their divisions. Chandler says this line-and-staff concept, in which line managers order the people doing the basic work and staff managers set standards, was first set out in American business by the Pennsylvania Railroad in December 1857.

What this means for teams today: McCallum's six principles still make a fair audit of any chart. Can each person name who they answer to, what they can decide, how the result gets checked, and how fast bad news travels?

The Rare Chart (1858-1914)

Railroads draw them, and most firms still don't

After the Civil War the template went to work on other rails. Chandler says the line-and-staff form was widely used, often in modified form, by roads such as the Michigan Central and the Chicago, Burlington and Quincy, and that nearly all American railroads carrying heavy traffic over long distances came to use it in time. British and European roads more often used a "departmental" structure, in which the functional managers in each division reported straight to their functional bosses at headquarters. The American roads wrote duties into manuals and charts, and large roads in the 1850s already employed forty to sixty full-time salaried managers. By the 1870s, Chandler's book prints a simplified chart of a large railroad, with full-time vice presidents heading the functional departments.

The scale kept growing. Harvard Business School's Baker Library notes that the Pennsylvania employed over 110,000 workers by 1891, more than the U.S. Army, Navy and Marines combined. The same page quotes the historian Richard White, who warned that railroad charts sometimes dissolved into networks of dependence, cronyism and kinship. The drawing was never the same as the practice.

Outside the railroads, the chart barely traveled. In 1914 the engineer Willard C. Brinton published Graphic Methods for Presenting Facts and wrote that "Organization charts are not nearly so widely used as they should be." He reproduced several, including one for a large stove-making company, and argued that a chart should be drawn for every business, because it "closely resembles a genealogical tree," with authority flowing down the branches. His reasoning was practical: with a chart, a worker would know whom to take orders from, and a general's order given straight to a private weakens the captain. He was blunt about the state of industry, writing that very few firms have an organization that can be charted at all.

What this means for teams today: A template travels when the people copying it have the same problem. The railroads copied the Erie because they all faced the same scale. If you borrow another company's structure, check that you share its problem first.

The Divisional Chart (1914-1929)

Engineers teach charting, and Du Pont and GM split the company into divisions

In March 1920, a five-man group at Du Pont, after six months of study into the company's marketing troubles, submitted a report that proposed, per Chandler, a new form of management structure. The group drew an organization chart showing the paints and varnishes line and the Pyralin and Celluloid articles line as "practically self-contained" units, each with its own purchasing, manufacturing, sales and accounting under one executive responsible for profit and performance. The proposal was rejected. But it shows the idea at the drawing stage: a chart whose boxes were products, not functions.

The ground had been prepared. In his General and Industrial Management, Fayol proposed a rule that every year, when the annual plan is drawn up, a scrupulous study of the organization be made "with the assistance of summarized charts." Such charts show each manager's immediate superior and subordinates, he wrote, and serve as a photograph of the framework at a given moment. Defects that are hard to see in a description leap into view on a chart, he added, and the charts also help avoid dual command. His wider case is in our history of Henri Fayol and the principles of management. For the shop-floor counterpart of this period, see the history of scientific management, and for the older question of why work gets split into boxes at all, the history of the division of labor.

Then came the crisis. At General Motors, Chandler writes, the 1920 collapse left inventories far above the committees' allotments, and William Durant retired as president on November 20, 1920. Ten days later Pierre du Pont took over. One of his first acts was to approve the plan Alfred Sloan had worked out months earlier, which the board approved on December 29, 1920. Durant had reviewed it and not acted. Sloan's plan created a general office to coordinate, appraise and set broad policy for the operating divisions, and Chandler says it remains the corporation's basic organization. At Du Pont, the Executive Committee approved a new plan on September 8, and in September 1921 the company put the structure into effect: autonomous, multidepartmental divisions under a general office with staff specialists.

Chandler's own account stresses that the executives at Du Pont, General Motors, Jersey Standard and Sears developed their structures independently, with no imitation. By 1925, he adds, GM had defined lines of authority and communication between the general office and its divisions, backed by a mass of accurate data. The chart had changed shape. The railroad tree asked who reported to whom for the day's trains. The divisional chart asked who owned a profit.

What this means for teams today: Every boxed unit on a chart answers a design question: what do we group by, function, product or place? Du Pont's rejected 1920 proposal and its adopted 1921 plan show that the answer can change within a year when the company changes.

The Span and the Shadow (1933-1960s)

Mathematicians count the lines; researchers find the people the chart omits

By the 1930s the chart had grown wide enough to worry about. Take the example Lyndall Urwick printed in the 1937 volume Papers on the Science of Administration: a committee reporting in 1930 on the League of Nations Secretariat proposed an arrangement with no fewer than fifteen independent officials reporting direct to the Secretary-General. Urwick used it to show what happens when the span of control is ignored, and he stated the standard of the day: no human brain should attempt to supervise directly more than five, or at most six, other individuals whose work is interrelated.

The mathematics came from V. A. Graicunas, a management consultant in Paris, whose paper from the March 1933 Bulletin of the International Management Institute was reprinted in the same volume. His argument was that a supervisor does not manage only single relationships with each subordinate, but also group relationships and cross relationships among them. Adding a fifth subordinate to a group of four, he calculated, raises the possible relationships in the unit from 44 to 100, a 127 percent increase in complexity for a 20 percent increase in working capacity. The number is a model, and Graicunas himself allowed an exception for routine work, where the subordinates' tasks do not interlock. But it gave the narrow-span rule a mathematical argument. What happens to that rule when software does part of the work is the subject of span of control when agents do the work.

The other challenge came from the shop floor. In Management and the Worker (1939), the Hawthorne researchers F. J. Roethlisberger and William J. Dickson wrote that too often it is assumed that a company's organization corresponds to a blueprint plan or organization chart. Their verdict: "Actually, it never does." The formal plan shows the functional relations between working units, they wrote, but not the primary groups of people in daily face-to-face contact. They also warned against treating informal organization as bad. It exists in every plant, they said, and much collaboration happens there. Our management theory timeline and schools of management thought show where that human-relations insight sits among the rival schools.

The clean boxes and lines of a chart suit an organization run by defined roles and written rules, and bureaucratic leadership covers both the strengths and the failure modes of that model. Where that model came from is the subject of our history of Max Weber and bureaucracy.

What this means for teams today: Count your spans before you draw your boxes. And compare the chart with where decisions are actually made. If the two differ, the chart is the one to fix, or at least to footnote.

The Chart as Data (1960s-present)

The diagram becomes a field in a database

Two escapes from the single line of authority arrived next. The matrix put two lines on one chart, and flat designs tried to remove layers. Both challenged the assumption that every person has one box under one box. We cover them in the history of matrix and flat organizations, and their more recent echo in org design and delayering.

The bigger change was in the medium. Walk into a company today and the org chart is rarely a sheet of paper. It is a record. In June 2006, RFC 4524 defined a standard "manager" attribute for LDAP directories, which "specifies managers, by distinguished name, of the person (or entity)." Each person's record points to the record of their manager, and the chart is whatever you get by following the pointers upward. Microsoft's directory API works in the same spirit: its documentation for the manager endpoint shows how to expand the chain of managers up to the root. McCallum's drawing was a picture of authority. Where approvals and permissions are built on a field like this, the record matters more than any picture.

That makes Fayol's annual audit cheap to run, and it exposes the old problem. Real work rarely follows a single manager line. Decisions run through informal groups, projects and shared tools that a manager field does not capture, which is Roethlisberger and Dickson's point in database form. One practical answer is to chart decisions as well as titles, with a tool such as the RACI matrix.

The open question is whether software agents get boxes. If an agent can approve a refund or send an email, the chart needs to say who answers for it, which is the question McCallum put to his superintendents in 1855.

What this means for teams today: Treat the chart as two things: a record in your systems and a picture people read. Keep the record accurate, because permissions and approvals depend on it, and keep a second view for how decisions actually run.

Key People

Key Facts

• The Erie's board asked in autumn 1853 for more precise accountability for expenses and a better appraisal of managers, and promoted McCallum to do it.
• Chandler says the Pennsylvania Railroad first set out the line-and-staff concept in American business in December 1857.
• Brinton wrote in 1914 that organization charts were not widely used as they should be.
• GM's board approved Sloan's plan on December 29, 1920, and Du Pont put its divisional structure into effect in September 1921.
• Graicunas calculated that a fifth direct report raises relationships in a unit from 44 to 100.
• RFC 4524 defines a directory attribute that records a person's manager.

The Pattern Behind the History

Our synthesis, not published research. The chart's history shows five repeated moves.

• Scale breaks sight. The Erie, the Pennsylvania and Du Pont each grew past what one person could watch.
• Information makes the lines. McCallum's telegraph reports and daily returns came with the drawing, not after it.
• A working template travels. The Erie's plan was copied by the Pennsylvania, and the Pennsylvania's by other roads.
• The picture omits the people. Roethlisberger and Dickson found the groups and norms the chart never shows.
• The medium absorbs the message. The drawing became a directory field, and the field now decides access.

What It Teaches Teams Today

• Audit the chart against McCallum's six principles. Check responsibilities, authority, verification, speed of bad news and daily information.
• Review it every year. Fayol's rule of an annual study with summarized charts is still cheap and still catches dual command.
• Count spans before adding boxes. Use Graicunas's model as a warning, then adjust for how interlocked the work is.
• Chart decisions as well as titles. A RACI matrix shows who decides what, which the boxes do not.
• Keep the record clean. If the manager field drives permissions, an out-of-date chart is a security and approval problem, not a cosmetic one.

The organization chart began as a railroad's answer to a line too long to see. It lasted because every large organization keeps running into the same problem. For the dated map around it, see the management theory timeline, and for who gets credit as the field's founders, see who is the father of modern management.