ence” consists in making small modifications in existing equipment and ob-
serving the empirical results, or in applying equipment to untouched areas of
observation. More revolutionary developments can be produced by inventing
new kinds of equipment, usually by analogy, recombination, or reapplication
of older equipment, deliberately cross-breeding equipment genealogies. The
telescope and microscope opened up new realms in which discoveries were
more or less guaranteed; the invention of the electrical battery in the late 1700s
and its application to electrolysis of fluids resulted in the discovery of dozens
of chemical elements, just as successive generations of particle accelerators were
responsible for producing the next round; the shift from optical to radio
astronomy widened the range of stellar phenomena, as did the combination of
spectroscopic analysis with optical photography (Price, 1986: 237–253). When
tinkering with the prior generation of research-front equipment becomes nor-
mal procedure, scientists come to expect new discoveries as a matter of routine.
The greatest attention goes to whoever is on the research forefront using
the latest equipment. New discoveries upstage older ones. The result is a social
consensus once a field stops being an area of hot research news. Scientists give
up competing over older theoretical rivalries so that they can move on to the
forefront of the newest, most successful equipment; they abandon old contro-
versies so they can get into new controversies. Of course not everyone aban-
dons old positions. Thomas Kuhn argued that proponents of defeated para-
digms never give up but only die out. It is likely that when the research front
is moving quite rapidly, adherents of older theories just as rapidly lose atten-
tion, and stubbornness does them little good. Thus the law of small numbers
may prevail on the forefront itself as it gives way to domination of one position
behind it, precisely because the lure of working on the forefront is much greater
than expounding an already known position.
There is another way in which research technology produces consensus.
When a new technology is tried out, the practical activity of scientific research
consists in tinkering with it to produce new phenomena, and then further
tinkering until those phenomena can be reliably repeated. This is not an easy
or automatic process. It took about 15 years for Boyle’s generation of air
pumps to produce consistent results; and today controversies swirl when cold
fusion or gravitational waves cannot be routinely evoked (Shapin and Schaffer,
1985: 274–276; more generally see Galison, 1987). The equipment—and the
theory—are perfected simultaneously when these effects are routinized, that is
to say, when the machinery embodies enough past practices of tinkering so
that one can follow explicit procedures and get expected results. This technical
repeatability is what makes science seem certain and hence objective.
Science is socially constructed, but I would stress that it is constructed not
so much by the imposition of ideas upon the world (idealist constructivism) as
Cross-Breeding Networks and Rapid-Discovery Science • 537