2 Evolution of knowledge in a world of increasing mechanization.

Machine tools, invented circa 1800, brought mechanical power and control to metal shaping. During the first three epochs of manufacturing, from 1800 to the early 20th century, the precision of these machines was progressively increased, mainly by mechanical means that constrained the behavior of machines and workers. The key developments of this period emphasized knowledge about different portions of the machining process (see Figure 1.5).

Little formal knowledge about any portion of the machining process existed prior to 1800. Quantitative measurement of parts not yet existing, the goal was to make each new firearm as similar as possible to the shop's working model. Even the conformance of finished parts to the model was judged idiosyncratically, by eye and caliper. Beyond this little can be said. Plates from Didier's Encyclopedia illustrate the range of hand tools available and undoubtedly there was qualitative knowledge (both verbal and tacit) about when and how to use them to achieve desired results.

2.1. English System

Different epochs emphasized the development of knowledge about different subsystems of processes. The state of technological knowledge in the English System is little documented, but we can infer general properties of the knowledge from what was achieved during that epoch. Technological breakthroughs revolved around three subsystems: the machine, specification of intended outcomes, and measurement of actual outcomes (Table 2.1).

Maudsley's achievement of highly accurate parts measurement using micrometers was accompanied by the invention of the engineering drawing. Accurate measurement and an absolute goal provided by the engineering drawing enabled a distinction between "better" and "worse" parts, which otherwise would have been judged merely "different" as in the Craft epoch. Taken together, the micrometer and the engineering drawing supported the creation of a basic feedback loop: keep removing material until a part is of the dimension specified in the drawings as measured by a micrometer. [15, Section 3]

Woodbury described Maudsley's other key contribution, the general purpose machine tool with highly precise lead screws for accurately cutting parts with a minimum of trial and error, in the four key elements: ample power and drive train sufficient to effect its delivery; adequate rigidity under the stress of cutting ferrous metal; precision in construction greater than the precision of the parts to be produced; and adjustability to accommodate flexibility in the parts. [44, pp 96-97] At a minimum, enough was thus known to design and build iron machines with these properties.

2.2. American System

The American System introduced new concepts of ideal outcomes based on tolerances and precision as well as accuracy. The corresponding new measurement method was the use of go/no-go gauges.

"Accuracy in this system, which might be as close as a thirty-second or sixty-fourth of an inch, was ensured by an elaborate system of patterns, guides, templates, gauges, and filing jigs." The use of these geometric devices to constrain the motion of cutting tools required the development of causal knowledge about linkages from jigs to final parts (Figure 2.1).

[FIGURE 2.1 OMITTED]

Colt and others developed, in parallel with knowledge about making firearms, the knowledge needed to design and build machine tools for specific purposes. Workers independent of those employed in the manufacture of firearms "built, maintained, set up, and improved machines." Specialized machine tool companies emerged to sell these machines abroad to furnish entire firearms factories.

Implicit in the emergence of these companies is another fundamental innovation of this epoch: separation of organizational knowledge by causal module. A machine tool designer does not need to know what parts are to be fabricated, only how to construct a machine capable of cutting along precise trajectories. The parts maker need not understand the nuances of how the machine works, only a limited range of adjustment methods. Information is transmitted from one to the other through the jigs. This separation of toolmakers' from tool users' knowledge is vital to the success of capital equipment industries.

What conditions support this separation of users and suppliers? There are two key conditions, one physical, the other having to do with knowledge.

First, the technology itself must have a modular causal network, that is, the total causal network must be separable into two subnetworks with much denser connections within than between them. The comparatively few connections between the subnetworks must be almost entirely in a single direction. Such a network structure is observed, for example, with geographically separated suppliers and customers between which there is a one-way flow of intermediate product. Causal paths tying the firms together pass through these intermediate products.

Second, knowledge about the causal relationships that join the subnetworks must be sufficiently complete to enable the modularity to be exploited. The key relationships that link the subnetworks must be well understood and their variables be known and measurable.

If both conditions are met, each subnetwork can be controlled by its own organization (department or firm) and the two joined by an arms length relationship. In Figure 2.1, a cutting tool's trajectory is a function of only a limited number of machine tool properties. Knowledge about the causal linkages among these properties was sufficient in the American System to make separate machine tool companies feasible.

2.3. Taylor System

Their extensive research on the "hard" technology of machining would render the impact of Taylor and his team on the transition from art to science fundamental, even in the absence of their more well known work at the Watertown Arsenal on worker procedures and standardized methods for each job. Conducted in secret for more than 20 years, the research was finally presented, in 1906, to an overflow audience of 3,000 at a gathering of the American Society of Mechanical Engineers. [35]

As in the other epochal shifts Taylor did not so much add to the established body of knowledge in its own terms, as shift the nature of the knowledge sought. His fundamental contributions to technological knowledge were several (see Table 2.3).

* Taylor's reductionist approach to systems analysis divided parts production into linked subsystems, each carefully analyzed in isolation to arrive at a formally specified "best" process. He studied not only parts machining, but also indirect supporting activities.

* Taylor moved from qualitative and ordinal relationships among variables to systems of equations with numerical coefficients that could be solved quantitatively.

* Finally, he employed a much superior learning method, namely a large number of carefully controlled empirical experiments, to develop knowledge systematically.

* These three contributions enabled Taylor's team to make specific discoveries about better manufacturing methods, perhaps most important their discovery of high-speed steel.

Each of Taylor's contributions constitutes a move from art towards science. The scientific knowledge he developed was a prerequisite for the development of standardized work procedures--his "one best way"--for which he is more famous. In Taylor's view, the best way could be determined only after the behavior of each subsystem was understood and had been quantified. Thus, for each subsystem, he moved towards science along the knowledge axis in advance of corresponding movement along the procedural axis. We consider these advances in turn.

2.3.1. Reductionist Approach to Manufacturing Systems

Taylor's insight was that production encompassed a host of distinct processes that could be analyzed and improved independently of the larger system they comprised. The sharpening of a tool, in his view, could be managed and optimized independently of the purpose for which the tool was to be used. As with the separation of capital equipment from firearms manufacture in the American System, this is feasible if and only if there is causal knowledge modularity. Taylor further realized that separation, analysis, and improvement could be applied to auxiliary processes such as accounting and maintenance as well as to materials processing.

Taylor applied this approach to all activities that had a significant effect on the overall rate of production, for example, the power transmission system (pink areas in Figure 1.5). The electrical motors of Taylor's day were large and expensive, so a few central motors powered dozens of machine tools by means of a network of moving belts. [15, Figure 5.1]

Inasmuch as the speed of operators was largely determined by the

speed of the machines as driven from a central location by belts,

pulleys, and shafts, Taylor considered the standardization and

control of these systems at their optimal level of efficiency

essential. To this end he established the activities of belt

maintenance and adjustment as a separate job and prescribed methods

for scientifically determining correct belt tensions. [15, Section

5]

A great deal of the old belting was replaced with new and in

some cases heavier belting. This made it possible to run machines

at higher speeds and with greater power, so that full advantage

could be taken of the cutting powers of high-speed steel, and also

prepared the way for Barth's later standardization of cutting speeds

and feeds. By the end of April 1910 the belt-maintenance system