An understanding of the history of manufacturing can aid our understanding of the systems we have today. Achieving part-to-part interchangeability and minimizing dimensional variation are often seen as key indicators of quality. But that hasn’t always been the case. In this article, I’ll explore how manufacturing has progressed from craft-based production to the modern lean-production systems. Along the way it will become clear how controlling variation has been fundamental to this development.
Traditional Craft-Based Production
Traditionally, goods were made by skilled craftsmen who were able to produce a complete product using highly adaptable tools. This is not to say that the importance of a division of labor in increasing efficiency has not been understood since antiquity. The Greek scholar Xenophon was the first known to have described it in about 400 BC. In his biography of the Persian ruler, Cyrus the Great, Xenophon stated that, “In a small city, the same man must make beds and chairs and ploughs and tables, and often build houses as well… It is impossible that a single man working at a dozen crafts can do them all well, but in the great cities… a single craft will suffice for a means of livelihood, and often enough even a single department of that; there are shoe-makers who will only make sandals for men and others only for women. Or one artisan will get his living merely by stitching shoes, another by cutting them out, a third by shaping the upper leathers, and a fourth will do nothing but fit the parts together. Necessarily the man who spends all his time and trouble on the smallest task will do that task the best.”
Despite the fact that there has always been some division of tasks within manufacturing processes, this was not sufficient to transition from craft-based production. There were some examples in the early industrial revolution where the production of simple products could be broken down in such a way that a number of unskilled workers carrying out highly repetitive tasks were able to produce products that would previously have required skilled craftsmen. This type of highly specialized division of labor is famously described in 18th century pin production by Adam Smith, although then it was seen that in more complex industries “the labor can neither be so much subdivided, nor reduced to so great a simplicity of operation,” (Smith, 1776).
Although the industrial revolution in Britain saw the automation of textile production and simple industrial items such as pins and nails, the manufacture of machinery continued to be carried out by craftsmen. The production of automobiles in Europe was still being carried out in a craft-based industry in the 1890s. One workshop would form steel parts to approximate gauges, harden them leading to distortions and then re-machine them. The parts produced in different workshops would be filed to fit to one another by skilled fitters. These craftsmen would file parts until they fit together. Each car was different, and its components could not fit to another car.
Early Examples of Interchangeability
At first, the fact that the parts of machines were individually fitted together didn’t appear to be a problem for the production system. It was, however, an issue for repairing military equipment. A number of early examples of interchangeable parts were focused on the logistics of supplying standard spare parts so that military hardware could be rapidly repaired in the field.
As early as the 10th century, Chinese crossbows were produced to standardized measurements. It has been suggested that this may have been so that interchangeable parts could be used to quickly repair them, although no records conclusively demonstrate this.
The earliest known example of standardized dimensions for the explicit purpose of interchangeability was the De Vallière system in 1732. This involved cannons that were cast in bronze and bored using a special boring machine. The improved tolerance of the bore fit a standard cannon ball, giving greater power and simplified logistics. This was still largely craft production, but the concept of interchangeability through working to gauges was born and subsequently developed further by Gribeauval in 1765.
Gunsmith Honoré Blanc took the idea of interchangeability a stage further, applying it to all the mechanical parts of a musket. He used master parts and gauges so that parts were filed and fitted to the gauge rather than to each other, making them interchangeable. In 1790, he demonstrated full interchangeability by manufacturing a thousand muskets and putting all their parts into separate bins. He arranged a demonstration in front of scientists, politicians and generals. He asked them to pick a part at random from each bin and assemble a musket from the parts. During the Napoleonic wars, Blanc was producing 10,000 muskets a year, although the French later moved away from interchangeability to maintain the skill of their craftsmen.
These early examples of interchangeability had great advantages in terms of logistics and field repairs. The military could carry standard spare parts and repair guns themselves rather than needing skilled craftsmen to repair guns. The system didn’t immediately lead to more efficient production systems. This was still an iterative process of filing, checking and filing again by a skilled craftsman. The only real difference was parts were fitted to a gauge rather than to another part.
Mechanized Production – Armory Practice or the ‘American System’
French arms manufacturers failed to realize the potential of their new system, but others were quick to do so. By 1803, the Portsmouth naval dockyard was mass producing interchangeable parts for pulleys and blocks for the British Navy.Marc Isambard Brunel, Henry Maudslay and others had achieved this using a series of machines, powered by a steam engine, arranged in lines to perform successive operations on the parts.
The machines included circular saws, bores, turning machines and morticing machines. They included a range of innovations that became standard features of mechanized production systems. The first machine in a sequence of operations would indent locating points that were used to consistently locate the parts in successive operations. Setting jigs and machine stops were used to produce cuts of the correct depth. Lead screws were used to control the advance of the workpiece into cutting machines. Lathes used tool holders and multi-jawed chucks.
Using these machines, laborers could produce over 10 times as many parts as skilled craftsmen had previously. However, these considerable achievements did not lead to widespread use in Britain. In fact, they were largely forgotten until after similar methods were developed in the United States and imported back into Britain.
Around the same time as the machines for pulley production were being developed in Portsmouth, Eli Whitney was attempting to mechanize musket production in the U.S. He had seen how the French were producing muskets with interchangeable parts and saw this as the key to enabling mechanized production. Getting a machine to fit parts to one another, as a skilled craftsman did, would have been very difficult. Getting a machine to fit parts to a standard gauge seemed achievable. He said that, "One of my primary objectives is to form tools so the tools themselves shall fashion the work and give to every part its just proportion.”
The key to producing parts within a variability that allows interchangeability in precision mechanisms was the development of the rational jig, fixture and gauging system. The principle of this system was that a physical model of the product was first created. A series of fixtures to hold the components in machine tools, jigs to guide operations and gauges to check the components were designed and built with reference to this model.
By 1819, the U.S. Federal armory in Springfield had a system in which workers had gauges to check parts during production and shop foremen and inspectors had gauges to check completed parts. These gauges were regularly checked for wear against master gauges or the model part. Each time a part was moved from one fixture to another, there was potential for errors to be introduced. In the 1820s, John Hall introduced a system of using a common reference point to locate the part in all fixtures to minimize these errors.
Parts for small arms production were typically produced by drop forging and then turning, milling, drilling and grinding. A range of special and single purpose machine tools were developed to carry out sequential operations. An important example is the Blanchard lathe used to shape gunstocks in the 1820s. This machine linked the rotation of two spindles, one spindle held a pattern of the gunstock and the other the blank to be shaped. As the spindles rotated, an arm carried a wheel over the pattern and a rotating shaper over the blank. In this way, the irregular shape of the gunstock was reproduced. The shaped gunstock was transferred to a sequence of special purpose machines, which created the recesses for the barrel, lock and trigger.
The new methods of production that were developed by small arms manufacturers were known as armory practice in North America and the American System in Europe. They were later adopted and further developed by producers of sewing machines and bicycles. Thebicycle manufacturers developed techniques for working with sheet metal, such as pressing, stamping and electric resistance welding. These were later to be widely used in the mass production of automobiles. The key innovations were the use of single-use machines used to carry out successive operations on parts and the use of jigs, fixtures and gauges to facilitate the interchangeability of parts.
Mass Production
Mass production was really an evolution that was based on armory practice. Before Ford introduced the Model T in 1908, he had first strictly enforced the principles of armory practice; working to gauge and using specialized tools arranged in the sequence of required operations. He intensively developed the design for manufacture of his cars, going through 20 different designs within five years. By the time he got to the Model T, it had been highly optimized for manufacturability with completely interchangeable parts that attached to one another simply.
The single innovation that is most closely associated with the birth of mass production is the continuously moving assembly line on which the car was moved past stationary workers who each added parts with small highly repetitive operations. Ford took the idea of performing simple operations in sequence in a line, which had been applied to mechanized part production, and applied it to assembly. In 1908, workers were still carrying out over eight hours of assembly, on fixed stations, before repeating an operation. By 1913, immediately before the moving line was introduced, this time was reduced to 2.3 minutes. The moving line reduced this to 1.2 minutes for the same operations. Without interchangeability this would have been impossible.
Assembly workers on this type of line generally do not require any particular skill and can be trained in a matter of minutes. In effect, the workers themselves are interchangeable. The disadvantage of this is that the workers cannot be expected to gauge parts to ensure quality standards are maintain or actively engage in the continuous improvement of the production system. Separate foremen and production engineers are required to carry out those functions.
In the 1950s, Toyota would show that it is possible to re-engage production workers within highly automated production systems. By introducing rapid tool changes and continuous improvement, they were able to reduce the need for separate foremen and production engineers, instead involving the workers in directly adding value to the product. These principles would later become known as lean manufacturing.
Product Specification Using Traceable Measurements
It is interesting to note that earlymechanized production systems had traceability of measurement only to a proprietary master part. This enabled interchangeability of parts produced in a single factory. It would not, however, enable standard parts to be used for multiple products within a global supply chain. Achieving a global system of traceability in which every part is produced using a gauge, which is traceable to a master part, would simply not work. Since different parts are used within different products, it would become impossible to identify which master should be used.
The solution was to make a dimensioned drawing of the product master. Tolerances can then be used to define what is required for parts to fit together. To ensure that all parts within a global supply chain are produced to the same standard, only the measurement equipment needs to be traceable. Rather than having a different master part for each product, we now have a single definition of the meter, which all drawings refer to. Factories all around the world calibrate their instruments with traceability to the fundamental definition of the meter, in terms of the speed of light. You may be thinking that your factory works in inches, not meters. However, since 1964, every country in the world has defined the inch as being exactly 25.4mm. It is, therefore, traceable to the international definition of the meter.
This issue was first understood with respect to threads. Standard threads were specified in terms of standard measurement units as early as 1800, when Henry Maudslay created the first screw cutting lathe and micrometer. Whitworth later created the first nationally agreed standard thread. However, it was not until much later that the industry saw the widespread use of dimensioned drawings together with multi-purpose gauges, traceable to internationally agreed standards of measurement.
When variable measurements are used to set machines and monitor product variation, it opens up many other possibilities. Simple go/no-go comparisons with gauges that are only calibrated against a master part provide little opportunity to understand trends in the process. When variable measurements are used, it opens up many possibilities such as the Statistical Process Control of the manufacturing system.