© Blackwell Publishers Ltd 2002.

Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 1JF, UK
and 350 Main Street, Malden, MA 02148, USA.
THE RESPONSE OF OLD TECHNOLOGY INCUMBENTS TO
TECHNOLOGICAL COMPETITION – DOES THE SAILING SHIP
EFFECT EXIST?
Jonx Hovrrrs
Aarhus School of Business
\ns+n\c+
This article investigates whether firms react to a radical technological substitution
threat by a deliberate acceleration of innovation in their existing technology – the
‘sailing ship effect’. There have been repeated claims that the effect has been sig-
nificant as a source of innovation (Foster, 1988; Rosenberg, 1976; Rothwell and
Zegveld, 1985; Utterback, 1996). Detailed reexamination of two cases thought to
be exemplars of the effect reveals that it existed in neither. It is suggested that the
characteristics of historical, technological substitution processes prompt misinter-
pretation based on superficial knowledge. Brief review of two other cases further
supports this position. It is argued that if the phenomenon occurs, it is likely to
be rare.
ix+nontc+iox
Creative Destruction and Three Firm Reactions to Innovation; Exit, Switch and the Sailing
Ship Effect
Creative destruction was Schumpeter’s term for the process by which innovation
destroys existing firms and their technologies (Schumpeter, 1943, p. 81). It includes
the process of substitution of a new technology for an existing technology for some
defined market. Schumpeter had nothing to say about the possible reaction of the
established firms to this process, but we know from work in the management area
that there sometimes is an active response to the threat of creative destruction
(Cooper and Schendel, 1988; Cooper and Smith, 1992; Foster, 1988). From this
literature we can identify three generic strategies of response to the process of sub-
stitution, which can be referred to as exit, switch (to the new technology) and the
sailing ship effect (the acceleration of innovation in the old technology in response to
the threat from the new). Before we analyse this last we should say something of the
other two.
‘Exit’ may of course be a forced outcome of creative destruction, via liquida-
tion. However, it is a strategic response if the incumbent firm anticipates problems
Address for reprints: John Howells, Associate Professor, Department of Organisation and Management,
Aarhus School of Business, Haslegaardsvej, 10 8210 – Aarhus V, Denmark ( Joh@asb.dk).
Journal of Management Studies 39:7 November 2002
0022-2380
from future innovation and elects to exit the threatened market early and to its
advantage over ‘forced’ exit.
The decision to ‘switch’ from the old to the new technology is particularly inter-
esting and has been the focus for the papers cited above, especially Cooper and
Smith (1992). This latter paper examines eight product lines that experienced sub-
stitution effects; these range from ball point versus ink pens, to diesel-electric versus
steam locomotives. Much of the analysis concerns the behaviour of 27 established
firms, selected by Cooper and Smith for their dominant market position in the old
technology. All of these entered the new technology, but few managed to establish
as dominant a position in the new technology as they had in the old. A diverse
range of problems faced those wishing to switch; these included the problems of
internal groups which recognised that the advancement of the new technology
threatened their expertise and power, and the problems of judging how the new
technology would develop and which old competencies could be retained and
which should be shed (Cooper and Smith, 1992).
Quality of Evidence for the Sailing Ship Effect
Cooper and Smith provide the most thorough exploration of response strategies
published in the management literature (they consulted over 250 secondary
sources). However, they nowhere comment on the third response strategy, the so-
called ‘sailing ship effect’, which can be defined as an acceleration of innovation in
the old technology in response to the threat of innovation in the new technology.
Some earlier management articles on response strategies do refer to the sailing
ship effect, but they provide little hard evidence of its occurrence; it seems to
be assumed, while the major focus is once again on the strategy of switch. For
example, Foster refers to it as a ‘typical response’ (Foster, 1988, p. 220) and
describes how sailing ship producers actively sought to improve sail in response to
the threat of steam. However, Foster does not give any sources for his assertion
and it is likely that he is drawing on the widespread, but incorrect interpretation
of Gilfillan, to be discussed later in this paper. A paper by Cooper and Schendel
reviewed seven examples of response to technological substitution threat and
claimed that firms fought back in two ways: by improving the old technology and
by switching to the new. To ‘fight back’ by improving the old technology is a
description of the sailing ship effect, but their paper is almost entirely a discussion
of the switch decision; they provide no evidence for, or against the existence of
the sailing ship effect (Cooper and Schendel, 1988, p. 249). The absence of any
reference to the effect in the later Cooper and Smith article (Cooper and Smith,
1992) is evidence of a kind for the absence, rarity, or at least of the effect’s
relative unimportance compared to the switch and exit response outcomes.
Claims for the effect rely on case study evidence. There is of course the case of
steam versus sail in the nineteenth century which gave the effect its name through
Gilfillan’s work (Gilfillan, 1935a) and which is referred to by Foster, Rosenberg,
and Rothwell and Zegveld.
There is also a very strong claim for the existence of the effect in the nineteenth
century alkali industry by Rothwell and Zegveld.
The alkali industry case also demonstrates the so-called ‘sailing ship effect’ . . .
it illustrates how established companies can become locked into existing tech-
888 ¡. novrrrs
© Blackwell Publishers Ltd 2002
nological trajectories. Rather than attempting to capitalise on the possibilities
offered by the emergence of a superior new substitute technology, they vigor-
ously defend their position through the accelerated improvement of the old
technology. (Rothwell and Zegveld, 1985, p. 41)
Rothwell and Zegveld not only assert the existence of the effect, but also that the
strategy of accelerating old technology innovation was adopted rather than the
superior strategy of ‘switch’ to a demonstrably superior technology. If the superi-
ority of the new technology really was known to the old technology firms, we have
an extraordinary case of economically ‘irrational’ behaviour. This case promises
to develop our understanding of when and why firms adopt the ‘sailing ship’
strategy.
A recent book by Utterback devotes a whole chapter to the invasion of stable
businesses by radical innovation and concludes that the sailing ship effect is a
general strategic response to threatening innovation (Utterback, 1996, chapter 7).
Of course the established players do not always sit back and watch their markets
disappear. Most fight back. The gas companies came back against the Edison
lamp . . . with the Welsbach mantle . . . there was nothing incremental about
that. Purveyors of established technologies often respond to an invasion of their
product market with redoubled creative effort that may lead to substantial
product improvement based on the same product architecture. (Utterback,
1996, p. 159)
Utterback’s view depends on two more nineteenth century cases: gas versus elec-
tric lighting, and mechanical versus harvested ice. The gas industry is a particu-
larly promising candidate because the Welsbach gas mantle was certainly a radical
innovation; it improved gas lighting efficiency by five times and for a time it was
not clear that electric lighting would prove the more effective and efficient light-
ing source (Bright, 1949, p. 126).
However, it does appear odd that the effect can occur in the sail, alkali, light-
ing and ice cases, but that examples of it were not reported in Cooper and Smith’s
review article. Although the Cooper and Smith cases are more recent, they are
not obviously and systematically different from the other cases. If any historical
trend is to be expected, it is surely that firms would have greater ability to respond
to invading innovation, and that response might be expected to include the sailing
ship effect.
Why the Sailing Ship Effect is Significant
So far a number of authors have claimed that the effect is widespread, but the
economic historian Rosenberg has best articulated an appreciation of the eco-
nomic significance of the effect if this is so.
The imminent threat to a firm’s profit margins which are presented by the rise
of a new competing technology seems often in history to have served as a more
effective agent in generating improvements in efficiency than the more diffuse
pressures of intra-industry competition. Indeed, such asymmetries may be an
important key to a better understanding of the workings of the competitive
nrsroxsr +o +rcnxorooic\r coxrr+i+iox 889
© Blackwell Publishers Ltd 2002
process, even though they find no place at present in formal economic theory.
(Rosenberg, 1976, p. 205)
Although Rosenberg is discussing operational efficiency rather than induced tech-
nological change, clearly if the sailing ship effect is widespread, it threatens to force
us to rethink the very nature of ‘competition’ in the economy. One of the strange
questions that would be raised if this were so, is why firms that accelerate tech-
nological change as a response to substitution competition did not do so before that
competition arose – why did not the normal competitive process between the old
technology firms ensure the adoption of any possible innovations in that technol-
ogy? We are forced to radical alternatives such as that they may have deliberately
and collectively suppressed potential innovation until the advent of a major threat.
Or that when they realized the extremity of the threat to their existence they were
somehow so strongly motivated that they were able to organize differently, or
search or work differently, such that they were able to develop genuine radical
innovation as a response to threat. If the effect is widespread, then perhaps either
the suppression of innovation is widespread, or perhaps firms are just not suffi-
ciently strongly motivated by normal competition to innovate until threatened with
extinction? Such possibilities do not fit well with available theory, but their dis-
cussion can be postponed until the widespread nature of the sailing ship effect is
established beyond doubt.
However, Rosenberg, like others, also seems somewhat dependent on the sail
versus steam case. He writes that the builders of sailing ships ‘responded to the
competition of iron and steam’ (Rosenberg, 1976, p. 205). Then in a later passage
he becomes more careful about the source of continued sailing ship innovation.
The sailing ship of the 1880s was far superior to its predecessor of 1850 or so,
and it seems plausible to attribute this improvement to the strong competition
of steam. Obviously one cannot assert this with authority, because we do not
know what the sailing ship of the 1880s would have been like in the absence of
such inter-technological competition. But it seems like a reasonable conjecture.
(Rosenberg, 1976, p. 206)
If the effect exists and is widespread, then it seems that it is significant for our
understanding of both the general firm strategic response to the threat of substi-
tuting radical innovation and the process of competition in the wider economy.
We have several strong claims for its existence in four specified cases, with the sail
versus steam case most often cited. We have Rosenberg’s observation that opera-
tional efficiencies as a result of substitution threats are widespread, yet he seems
to prevaricate over the sailing ship effect’s existence in the exemplifying case of
sail. In the management literature there are no additional developed cases of the
effect, and a peculiar absence of mention of the effect in Cooper and Smith’s thor-
ough review of eight substitution cases, which is exactly where you would expect
to find further evidence. We may conclude at this point that the sailing ship effect
is surely worthy of further investigation.
Structure and Method of this Paper
The approach taken in this paper is to reinvestigate the basis for the strong claims
for the effect’s existence in the sail and alkali cases. If these cannot be supported,
890 ¡. novrrrs
© Blackwell Publishers Ltd 2002
and if more plausible and conventional explanations for the appearance of the effect
can be found, we will not have to revise our fundamental concepts of competition
and strategy. What now follows is a detailed analysis of the competitive process in
first the sail and then the alkali case.
The sail case has been chosen because it is so frequently cited that it has given
its name to the effect and it is also an example of a ‘typical’ nineteenth century
pattern of incremental and long term technological substitution: the ice, sail and
gas cases all have invading technologies which evolved through many incremental
innovation steps, and the rate of substitution was limited by this slow rate of
improvement of the new technology. The alkali case has been chosen because it
demonstrates a quite different pattern of substitution. The novel Solvay chemical
process technology was from its inception recognized as a radical and superior
process to the old Leblanc method of making sodium carbonate. Only the rate of
build of new plant obviously limited the rate of substitution – in this case above
all others, the Leblanc producers must have been aware of the threat of extinc-
tion and should have had time to respond.
It will become evident that neither the sail nor alkali cases demonstrate con-
vincing evidence of the sailing ship effect as an acceleration of innovation in the old
technology in response to the threat of the new. Therefore, in these two cases there
will also be a concern to establish how the substitution process could have been
interpreted by the authors concerned as evidence of the sailing ship effect. This
will be followed by an application of the lessons from the sail and alkali cases to
the ice and gas cases. Finally, there is some discussion of what this implies for our
understanding of the concept of technological competition.
+nr s\irixo snir rrrrc+ ix +nr s\irixo snir ixnts+nv
Although both Rothwell and Zegveld, and Rosenberg cite Gilfillan’s study of inno-
vation in sailing ships as the key example of the sailing ship effect, Gilfillan’s work
does not provide an unequivocal demonstration of the effect. Gilfillan’s work con-
sists of a survey of innovation in ships from earliest history (Gilfillan, 1935a) from
which he tries to draw general principles governing innovation in his ‘Sociology
of Invention’ (Gilfillan, 1935b). The ‘sailing ship effect’ is not one of Gilfillan’s key
principles, but relevant passages appear when he describes the prospects for sail
in the 1930s.
Only today, faced by the imminent extinction of all wind-blown vessels, do we
see signs of an impending revolutionary period – such signs as the auxiliary
engine sailer, the sailing barge, and especially since the War, the metal sail with
two stream-lined surfaces, and the rotorship. (Gilfillan, 1935b, p. 18)
Today the very threat of extinction seems to have stimulated greater inventive-
ness than ever . . . (Gilfillan, 1935b, p. 163)
But Gilfillan nowhere stipulates the motivation or the contribution of the sailing
ship producers to the increased effort at innovation. The nearest he comes is in the
following passage when he discusses the last shippers to build sailing ships in
Europe, in Hamburg in the late 1920s (Gilfillan, 1935b, p. 163).
nrsroxsr +o +rcnxorooic\r coxrr+i+iox 891
© Blackwell Publishers Ltd 2002
The decline of the sailing ship . . . had produced a feeling of crisis, of necessity
for action, among not only those directly concerned, but also by the steamship
people who valued the windjammers as a training school . . . (Gilfillan, 1935b,
p. 2l2)
Sailing ship producers are included among those who feel the industry has come
to a terminal crisis, but neither here nor elsewhere are specific innovations causally
associated with their sense of crisis. Which is not to say that they did not occur,
but Gilfillan provides scattered evidence of state intervention and support rather
than private sector investment driven by a sense of crisis.
So Gilfillan associates the sense of crisis within sail with the founding of the
Institute voor Aero-en-Hydro-Dynamik in Amsterdam, a research institute with
the object of studying the physical flow laws and their industrial application. The
established German inventor-entrepreneur Anton Flettner cooperated in creating
and then managing the institute; Gilfillan quotes Flettner,
The wind propulsion of ships attracted our interest more than anything else.
There was no doubt in our minds that the visibly progressing decay of the sailing
ship, which could not appreciably be delayed by the advent of the auxiliary
(motored) sailing ship, was not a fact which one had to accept like a law of
nature. (Flettner, 1926; cited in Gilfillan, 1935b, p. 2l2)
This is a clear expression of the belief that accelerated innovation in sail was
possible. However, it tells us nothing about the sailing ship effect, since we do
not know the contribution of private sector, old technology firms to the research.
It cannot have been very much, since in 1926, only a few years after the founda-
tion of the institute in 1922 (Gilfillan, 1935b, p. 212) ‘building had practically
ceased’ (Gilfillan, 1935b, p. 163).
Gilfillan and Flettners’ enthusiasm for the renewed future of sail depended
largely on the prospects for Flettner’s ‘rotor’ ship. What is important to our dis-
cussion is that the second and most important prototype of the rotor ship was paid
for, not by a sailing ship firm, but by the German Navy. By 1935, even the opti-
mistic Gilfillan comments that the rotor ship appeared to have run into economic
difficulties. More than 60 years later, we know that neither the failed rotor ship
nor any of the other projected radical innovations saved the sailing ship in its core
markets of freight transportation.
Not only does Gilfillan not make an explicit claim for the ‘sailing ship effect’ as
a private sector phenomenon, he provides persuasive evidence that government
was the more important funding agency for these attempts at radical invention.
Does this mean the sailing ship effect is an imaginative fiction? To complete the
analysis we must look at the pattern of real incremental innovations that changed
the sailing ship during the period of its displacement by steam. It will become clear
that it is far from clear that such innovation was ‘driven’ by competition from
steam.
The Record of Innovation in Sail
Much of the clearly important innovation in both steam and sailing ships was
derived from the use of cheaper iron, then steel. This allowed progressively larger
and therefore more economical hulls to be built for both steam and sailing ships.
892 ¡. novrrrs
© Blackwell Publishers Ltd 2002
In the case of sailing ships, first, wooden hulls were built around a steel frame,
later, hulls and sail masts could be made entirely of steel (Gilfillan, 1935b, p. 160).
Stronger ships and steel masts allowed greater areas of sail and greater wind power
to be captured. This source of economic advantage explains the continually
decreasing costs of sail freight contracts recorded at the Baltic exchange (Harley,
1971) and somewhat obscures the real substitution process of steam as a motive
force for sail.
It is clear enough that there is no need to postulate an increased sense of threat
to explain continuing innovation of this sort; the usual competitive process is suf-
ficient, where firms with similar technologies seek to better meet the needs of users
through the improvement of their products and services – in this case through the
exploitation of cheapening new materials.
In the search for understanding of the competition between the two motive tech-
nologies and in the absence of direct evidence of management intentions linked
to specific sail innovations, the best we can do is to reconstruct the steps in the
pattern of substitution.
The most obvious feature of this process was that substitution was hampered
by the slow evolution of steam as a power source. Early steam ships were disad-
vantaged because England was the primary coal source during the nineteenth
century and they had to carry their own fuel. As a consequence, steam cargo rates
rose dramatically with distance from England – the greater the journey from
England, the more cargo space had to be allocated to coal storage. This effect was
weakened as first coaling stations were established on major trade routes and then
eventually new sources of coal developed outside England and Europe.
Therefore the pattern of substitution was for first short, then increasingly long dis-
tance voyages from England to be taken by steam ships. This pattern was driven
throughout the period of coexistence of the two motive technologies by constant
improvements to the efficiency of marine engines and boilers (boiler operating
pressure was crucial and depended on techniques of metal manufacture).
What we have called the ‘period of substitution’ lasted for over 70 years and in
this time the volume of shipping grew enormously, so that although sailing ships
declined as a proportion of ships built (in the late 1860s two thirds of tonnage
built was sail; by 1872 it was only 15 per cent; Harley, 1971, p. 224), the largest
tonnage of sailing ships launched in a single year in Britain was launched in 1892
(Harley, 1971). Relative decline was masked by absolute growth in demand for
many decades and this effect, common to many substitution cases, might be
expected to mask the ‘threat’ from steam.
The market for shipping services also became increasingly segmented over this
period. The principal service advantage of steam over sail lay in its speed and reli-
ability, which is why passenger services were taken by steam as early as the 1850s
and 1860s (Harley, 1971). For the same reason, cargoes consisting of perishable
and high value goods like tea switched to steamers first. However, sail remained
economic compared to steam into the twentieth century for the longest voyages
to the pacific coast of the USA and to Australia, aided by the ‘economics’ of trade
winds, and for the shipping of bulk cargoes, like coal, of low tonne value (Harley,
1971). Gilfillan describes the design of sailing ships adapted to this market niche,
Their response was natural, cutting their crews, by abolishing the lightest sails,
broadening the upper yard . . . and cutting in two the largest sails until furling
nrsroxsr +o +rcnxorooic\r coxrr+i+iox 893
© Blackwell Publishers Ltd 2002
could replace almost all reefing. Sails and mast heights were standardised, and
more use made of the easily handled fore-and-aft sails, so that the great steel
four-mast bark, like the noble Brilliant, became almost standard around 1900,
the last established type. (Gilfillan, 1935b, p. 160)
Other adaptations were made, such as auxiliary steam engines to further reduce
labour costs and to increase manoeuvrability, for example in harbour.
As far as the sailing ship effect is concerned, our problem is that these opera-
tional adaptations were not necessarily made ‘in competition’ with steam, i.e. as a
conscious strategic response to declining costs of steam transport. They might just
as well have been adaptations to the physical characteristics of long distance
voyages, driven by competition between rival sailing ship firms for contracts. On
the very long voyages which comprised the remaining ‘market’ for sailing ships,
the listed innovations aided the efficient exploitation of trade winds, which pro-
vided reliable motive force; in deep seas and with unchanging, constant winds,
small crews became possible.
The Lack of Evidence for the Effect
Throughout this detailed account of the substitution process, we have this problem
that continuing old technology innovation is not evidence of the sailing ship effect,
for it may just as well have been driven by ‘normal’ competition between old tech-
nology incumbents. The evidence of clear radical invention like the rotor ship
(which failed) was financed and driven by government, not the firms supposedly
experiencing a great threat. There is no ‘innovation frequency chart’ to demon-
strate an increase in the rate of innovation in response to steam; all we have is
Harley’s chart of a monotonic decline in sailing ship freight costs, which if it is
anything is evidence against the sailing ship effect.
Decisive evidence might exist in primary company sources and would consist of
management statements that innovative projects were successfully carried through
as a response to steam. The only examples we have of such statements are asso-
ciated with, once again, the Flettner rotor ship effort, a government-inspired
project, but it remains possible that a thorough search of specialized library
archives in, perhaps, Hamburg, might produce this kind of evidence.
Given this situation it is still possible to discuss the characteristics of the
substitution process as they would be likely to impact on the sailing ship effect
‘decision’.
It is Unlikely that a Strong Threat was Perceived Throughout the Period of Substitution
The plausibility of the sailing ship effect hinges first on the perception on the part
of sailing ship producers that there was a strong threat to their existence. In order
to think about this, it may be useful to think of the two technologies as dominant
in two separate markets, but alternatives to each other only at some geographical
‘boundary’ between these markets. Over a period of more than half a century this
boundary moved to eliminate the sailing ship – but at any point in time the bulk
of orders for sail and steam ships would be for a market that was not imminently
threatened by substitution. The long term and generally incremental nature of the
substitution process and the rise in total market size for both sail and steam until
the mid-1890s suggest that for much of this period the ‘perceived substitution
threat’ on the part of sailing ship producers would have been relatively weak. The
894 ¡. novrrrs
© Blackwell Publishers Ltd 2002
only evidence we have that producers experienced a strong sense of threat is at
the very end of the substitution period (Gilfillan, 1935b) and by this time it was
apparently too late for producers to act.
It is Likely that the Strategic Alternatives of ‘Switch’ and ‘Exit’ Would Be Compared to the
(Possible) Option of the Sailing Ship Effect
One last issue pertinent to the sailing ship effect may be raised in this case; the
strength of the sailing ship builders’ commitment to the old technology, or the
degree of technological ‘lock-in’. We would like to know how difficult it was to
switch between the manufacture of sailing ships and steam ships. If it were easy
to switch from the old to the new technology, it would be a matter of relative indif-
ference to producers if the old technology were eliminated – they would simply
move to the new technology.
We do have some piecemeal evidence that the sailing ship producers were not
strongly locked-in to the old technology; in the mid-nineteenth century the same
producers also equipped early (and unreliable) steamships with sail (for example,
Brunel’s Great Western and Eastern steamships; Rolt, 1957). We also recall that
by the early twentieth century, sailing ship hulls were made from steel as steamships
were – sailing ship producers were acquiring some of the technological abilities
that steamship builders would need. All this may be taken to suggest that the
degree of lock-in to the old technology was certainly not insurmountable, and we
may therefore doubt that when it was finally recognized that sailing ships were on
the point of elimination there would have been a strong desire to react in their
defence (whether or not possible).
In general, we might expect that the effort of the sailing ship effect, even if pos-
sible, would – or should – be compared to the alternative strategies of ‘switch’ to
the new technology – and of course, the ‘strategy’ of exit to an entirely different
activity. In conjunction with the observation that ‘normal’ competition within the
old technology should result in adoption of old technology innovations, which
implies that the potential for increasing the rate of innovation in the old technol-
ogy is low to non-existent, we can suspect that the sailing ship effect, if it exists at
all, is not widespread.
We have a strong case for the absence of the sailing ship effect in the case that
above all others was thought to exemplify the effect. We turn now to the other
case that has been interpreted as providing strong evidence of the effect, the case
of rival alkali production technologies in late nineteenth century England. Here
we have a case where a strong sense of threat was experienced by the old tech-
nology firms and yet the strategy of ‘switch’ was blocked by the patent position of
the firm controlling the new technology.
+rcnxorooic\r coxrr+i+iox nr+vrrx +nr sor\v \xn rrnr\xc
rnocrssrs ron \rk\ri rnontc+iox ix +nr xixr+rrx+n crx+tnv
As in the sail case, the rival processes for producing alkali (sodium carbonate) co-
existed for decades. Unlike sail, alkali was an undifferentiated commodity product
sold into a national market. If the Solvay process was clearly superior to the
Leblanc, one would expect that the competition between the two processes should
be especially intense and resolved relatively quickly – only limited by the time taken
nrsroxsr +o +rcnxorooic\r coxrr+i+iox 895
© Blackwell Publishers Ltd 2002
to build new Solvay plant – or unless there was a sailing ship effect. Since Brunner
Mond obtained a licence for the Solvay process and began production in Britain
in 1872, but the last Leblanc plant was only closed in the first world war (Lindert
and Trace, 1971), there was an apparent protracted 42 year period of coexistence
of the two technologies. This begs for an explanation and with no further infor-
mation there is a tendency among some of the secondary sources to assume a suc-
cessful (in the sense of some economic gain) sailing ship strategy on the part of
the Leblanc producers; for example, Yu, below.
Despite the obvious advantages of the Solvay process, British managers, who had a large
investment in Leblanc plant, were unwilling to abandon it [ Yu’s emphasis]. Instead, they
responded to the challenge in a number of ways: by increasing efficiency
through process improvement; by concentration through the merging of com-
panies to form the United Alkali Company, thus facilitating economies of scale;
and through entering into price fixing agreements with British Solvay produc-
ers ( Yu, 1982; cited in Rothwell and Zegveld, 1985, p. 41).
The unexplanned mystery is why Leblanc producers were prepared to do all this
when the better available strategy was to switch to Solvay production. Like Yu,
Rothwell and Zegveld also stress the apparent irrationality of the Leblanc pro-
ducers, as does a third source, Lindert and Trace (although they have nothing to
say about the sailing ship effect; Lindert and Trace, 1971). The long period of
coexistence of the two processes when one is obviously superior to the other is a
problem requiring explanation and superficially it does appear to support the con-
clusion of irrationality on the part of the Leblanc producers.
However, it will be shown that an alternative and better explanation is available
for the adherence to the old technology that involves no irrationality and no sailing
ship effect, but only poor judgement. As with sail and steam, a more detailed analy-
sis of the case is required.
From Creative Destruction in a Free Market to a Rigged Market
Brunner Mond had a preferential license in Britain for the Solvay process, the
superior and soon dominant ammonia-soda method of sodium carbonate manu-
facture. The conditions of the patent were such as to make Brunner Mond the
effective monopoly supplier of Solvay ammonia-soda in Britain. Production began
in halting fashion in 1874 (Reader, 1970, p. 53).
Once Brunner Mond production became significant, Leblanc soda production
began a rapid decrease, despite the rapidly growing market for soda ash through-
out the second half of the nineteenth century. Figures compiled (Table I) from a
standard chemical industry history (Haber, 1958) show creative destruction well
underway in the 1880s, and crude extrapolation of trends in Table I suggest that
it would have been possible for creative destruction to have been completed before
1900.
Instead there is a dramatic cessation of expansion of Solvay production by
Brunner Mond around 1890 and static output for eight years. Brunner Mond had
entered a market fixing agreement with the Leblanc producers despite the very
evident success of the Solvay process against the old technology. This
was clearly in the Leblanc producers interests – it helped prolong the life of the
Leblanc system and is a partial explanation of the prolonged period of coexist-
896 ¡. novrrrs
© Blackwell Publishers Ltd 2002
ence of the rival technologies. However, it is now not clear why Brunner Mond,
with the proven superior technology, should want to enter an agreement with the
‘loser’ firms. To understand this firm behaviour, we need to know more detail of
the Leblanc and Solvay processess and the events behind their substitution.
Solvay as a Partial Technological Substitute for Leblanc Output
The key to understanding this case is that it is not a simple one of substitution
of one process for sodium carbonate production by another, as has been assumed
by some of the secondary sources (in particular, Rothwell and Zegveld). Both
processes produced byproducts and wastes whose economic value changed
through this period; as in the case of sail, we have to be careful about what ‘com-
petes’ with what. It will be argued that in the detail of such changes lies the expla-
nation of the Leblanc producers’ apparently ‘irrational’ adherence to the Leblanc
process.
The Leblanc process had more wastes and byproducts than the Solvay process.
It began with the treatment of salt with sulphuric acid, which produced hydro-
chloric acid. This was originally discarded, but by the second half of the nine-
teenth century it could be used as a key input for the manufacture of bleaching
powder, for which there was a new and growing market. Later, chlorine could be
extracted for another distinct market via the Deacon process.
The second stage of the Leblanc process involved the reduction of sodium
sulphate with limestone and coal or coke to produce either the carbonate or the
hydroxide (caustic soda) and ‘tank waste’, which was mostly calcium sulphide. So
caustic soda could be made at the expense of soda production and was more cheaply
nrsroxsr +o +rcnxorooic\r coxrr+i+iox 897
© Blackwell Publishers Ltd 2002
Table I. Solvay versus Leblanc production of sodium carbonate in
Britain in thousands of tons
Solvay (ktons) Leblanc (ktons)
(Haber, 1958, p. 158) (Haber, 1958, p. 152)
1878 4 1878 196.9
1880 18.8 1880 266.1
1882 233.2
1884 204.1
1885 77.5
1886 165.9
1890 179.5
1898 181
1900 225
1903 240
Data derived from Haber (1958, pp. 152, 158). Additional data: 1878
Solvay figure from Reader (1970, p. 53) and 1900 figure from Reader (1970,
p. 174). NB: The UAC refused to publish precise Leblanc soda output
figures from 1893 (Lindert and Trace, 1971, p. 282), while UK government
statistics do not distinguish between the output of Solvay and Leblanc
processes. Output figures later than this date are difficult to establish, espe-
cially as the alkali industry continued to differentiate into caustic, bleach-
ing powder and chlorine products as well as sodium carbonate. This table
represents a compilation of all output data which specifies the producing technol-
ogy that I have been able to find in the sources listed.
made via the Leblanc process than the Solvay (Lindert and Trace, 1971, p. 251).
As demand for caustic grew, the Leblanc producer deliberately switched produc-
tion from soda, so reducing losses and boosting profits at the same time (Haber,
1958, p. 96).
However, whenever bleaching powder was made, either sodium carbonate or
caustic soda would have to be made. This meant that despite the growth of these
new markets, the Leblanc producers would retain a limited interest in the state of
the sodium carbonate market.
The Logic of the Market Agreement
Prices for soda ash in Europe fell over 75 per cent between the 1860s and late
1880s (Reader, 1970, p. 54), a scale of decline greater than any estimated cost dif-
ference between the two soda production processes. With a general trade depres-
sion between 1884 and 1889, the decline accelerated, and while this aggravated
losses for the many Leblanc producers, it also brought near zero profit for Brunner
Mond by 1887 (Lindert and Trace, 1971, p. 257). It is this collapse of profitabil-
ity for all producers that induced the restructuring of the alkali industry and the col-
lective rigging of output and prices of the main alkali products.
The more than 40 independent Leblanc producers reacted to the slump in
prices by merging in 1891 into the United Alkali Company (UAC) (Reader, 1970,
p. 106). Haber writes of the promoters of the UAC, that they ‘made it clear they
intended to effect far-reaching reforms’ and that they were motivated to ‘establish
friendly relations with the Solvay manufacturers’ (Haber, 1958, p. 182).
These ‘friendly relations’ comprised an ‘agreement on price levels and market
shares for each soda product’ (Lindert and Trace, 1971, p. 253) and by its peri-
odic renegotiation and renewal the two parties maintained the rigged market into
the first world war.
Irrespective of the details of the agreement or its effective prosecution, on cir-
culation of rumours of the formation of the UAC, alkali prices began a specula-
tive rise, which restored and maintained overall profitability for both processes until
at least 1897 (Lindert and Trace, 1971). However, the details of the agreement do
reveal the strategic concerns of the two producers and in particular provide some
explanation for the otherwise astonishing entry of Brunner Mond into the market
agreement when it possessed the superior production technology.
Brunner Mond’s great vulnerability was the expiry of its major Solvay patent
in 1886 (Reader, 1970, p. 106). and with the creation of the giant UAC (1891)
a full or partial switch from Leblanc into ammonia soda production was at least
a theoretical possibility. In the 1891 agreement Brunner Mond obtained an agreed
upper limit for UAC ammonia-soda production of 15,000 tons/year. In return,
Brunner Mond agreed an upper limit to its soda production of 165,000 tons/year,
to limit production of chlorine and not to enter the market for caustic (Reader,
1970, p. 109). UAC subsequently made a limited entry into ammonia-soda tech-
nology with the purchase of two ammonia-soda plant in 1893 (Reader, 1970,
p. 108). The agreement therefore effectively ended creative destruction in the
carbonate market, but by limiting UAC entry into ammonia-soda production
preserved the identity of each of the firms with the major ‘rival’ technologies.
Brunner Mond had further reason to enter an agreement since Solvay soda
was only a partial technological substitute for the Leblanc process, which meant that
898 ¡. novrrrs
© Blackwell Publishers Ltd 2002
Brunner Mond could not expect to achieve the full creative destruction of the
Leblanc producers through output expansion. Such a policy risked the worst of
all worlds: a low sodium carbonate price and the danger of retaliatory entry into
Solvay technology financed by the premium prices obtainable in UAC’s mono-
poly over the bleaching powder and caustic soda markets.
Nor was an aggressive switch to ammonia-soda an obvious and profitable strat-
egy for the UAC. By 1891 Brunner Mond held over 50 per cent of British car-
bonate production (Table I). Major entry into ammonia-soda risked a further price
collapse and war with Brunner Mond. But if it came to a price war, Brunner Mond
had capital reserves while UAC was loaded with debt (Haber, 1958, p. 159). One
last deterrent faced a UAC strategy of massive entry into ammonia-soda – the
technical difficulty of operating ammonia-soda plant without Solvay group assis-
tance. The Solvay group collected details on individual plant performance at fort-
nightly intervals and in return, dispersed technical assistance and new patents
within the group. Reader comments on the development of the Solvay process
that,
. . . it will be evident that the Solvays’ success, when at length it came, depended
comparatively little on the kind of knowledge which is revealed in a patent
specification or expounded in a textbook. It depended far more on detailed
experience of the process in action and on trade secrets. These secrets were
and are very carefully guarded indeed . . . (Reader, 1970, p. 45)
One of the radical features of the Solvay process was that it operated continu-
ously and much of the difficulty of management was connected to this feature.
Given the years it had taken the Solvay brothers to build up this knowledge
and given Brunner Mond’ s powerful lead in operating the technology, UAC’s
strategy of negotiating a rigged market, maintaining a division of control of the
technologies and placing their hopes on the future of the chlorine and caustic
markets looks reasonable. In essence, by the time UAC was created and could pose
as a threat to Brunner Mond, it was too late for a strategy of wholesale switch into
ammonia-soda.
The Reaction of the Leblanc Producers to Solvay Production – Dubious Evidence of the
Sailing Ship Effect
The significance of these events for the Leblanc system is that it bought time and
increased profits with which the Leblanc producers might have sought to improve
their position. They had a strong motivation to act, combined with what appeared
to be time and resource – what did they in fact do?
There were improvements to the Leblanc system, but the greatest technological
innovation of the period of substitution came before the formation of the UAC and
was the Claus-Chance process of 1882, which made it possible to recover the
sulphur previously lost in the form of calcium sulphide (Haber, 1958, p. 100). This
could be recycled to the production of sulphuric acid, which itself had a growing
diversity of uses through this period.
We have no clear evidence that the development of the Claus-Chance process
was driven by the threat from Solvay production. Rather, the chemical histories
tell us that there had been many attempts over decades, and before the arrival of
nrsroxsr +o +rcnxorooic\r coxrr+i+iox 899
© Blackwell Publishers Ltd 2002
the Solvay process, to recover the valuable sulphur lost in the Leblanc process.
Texts like Haber comment only that theoretical cost savings and potential new
markets always provided the incentives to innovate and improve the Leblanc
system. As in the case of sail, continuing old technology innovation does not imply
the existence of the sailing ship effect.
However, the main sources cite UAC’s claims at the time of its formation that
it would ‘modernize’ and ‘reform’ its processes, which at first appears promising.
So the UAC claimed that its creation subjected the constituent Leblanc manufac-
turing plant to central comparison and control and this allowed it to establish best
operating practice in every works (UAC, 1907, p. 41). This is another example of
that ‘tightening’ of operational efficiency that Rosenberg noted as common in sub-
stitution cases and that we may have seen in the sail case. This does not really count
as the sailing ship effect, which is properly the induction of technological change –
involving change in the old artefact forms.
The UAC did found a ‘Central Research Laboratory’, which it claimed pro-
duced improvements in old processes (UAC, 1907, p. 41). If we knew what these
were, we would have examples of the sailing ship effect, but they are nowhere
specified in published company documents (for example, UAC, 1907) or the chemi-
cal histories. If they existed, they are unlikely to have been of great economic
importance, as a description of subsequent events will show.
Other UAC activities that are described are vertical integration into salt and
pyrites mines, the latter to secure sulphur supplies. There was also some diversifi-
cation – UAC bought a railway in Spain (Reader, 1970; UAC, 1907), which if it
suggests any strategy, suggests increased effort to exit Leblanc production.
The problem with the actions listed above is that while some are efforts to
improve the Leblanc system of production, they are not necessarily efforts to improve
the economics of sodium carbonate production. We defined the sailing ship effect as if two
substitutable products would compete and on this basis, for the sailing ship effect
to exist we would require evidence of efforts to accelerate innovation specifically
in Leblanc soda production. Of this we have no evidence. Instead the efforts at
improving operational performance were primarily intended to benefit the devel-
opment of the bleaching powder, caustic and sulphuric acid markets and appear
more as strategies of ‘exit’ rather than the sailing ship effect. Improvements in the
economics of soda production did occur, but they became incidental to the main
strategy of exit from soda production – they were never expected to enable serious
price ‘competition’ with the Solvay process.
The UAC did not publish output figures after 1893 (Lindert and Trace, 1971,
p. 282), but by their own published comments it appears that by 1902 at the latest
the UAC had ceased to make carbonate altogether (Lindert and Trace, 1971, p.
249; UAC, 1907, p. 33). It is reasonable to assume that some time earlier, perhaps
as early as 1890, they came to define themselves as primarily caustic and bleach-
ing powder producers, who happened to make some high cost sodium carbonate
as a byproduct of these operations.
It is now clear that the ‘paradox’ of the long period of coexistence of Solvay
and Leblanc production in Britain is no paradox at all. The ‘late’ (1890s) Leblanc
system was not primarily a sodium carbonate production technology as it had once
been, but a chlorine and caustic soda technology. Nor can the UAC be easily
judged as irrational by its decision to restrict itself to the development of these
markets – they appeared to be secure from Solvay competition.
900 ¡. novrrrs
© Blackwell Publishers Ltd 2002
Electrolytic Production of Alkali – Extinction of Leblanc
Unfortunately for the UAC this situation did not last long because new sources of
technological competition eroded its monopoly position in chlorine and caustic soda
product markets. Although Brunner Mond developed a means of competing in the
chlorine and caustic markets and used these to renegotiate the alkali market sharing
agreements in its favour (Haber, 1958, p. 158), the decisive blow was the develop-
ment of modern electrolytic production methods. Electrolysis of strong brine solu-
tion enabled the direct production of caustic soda and chlorine well below Leblanc
costs. Electrolytic production escalated from 1895 and rapidly eroded the remain-
ing profitable Leblanc markets (Haber, 1958, p. 160). The formation of the UAC
and the Brunner Mond market sharing agreement proved to provide a very short
reprieve for the Leblanc system – so short that it casts further doubt on the UAC’s
ability to deliver any real improvements to the Leblanc system. Once electrolytic
production began to affect market prices, it rapidly became too late for the Leblanc
producers to act. As Haber describes the period of precipitous decline,
The board soon discovered that the discouraging trading results prevented them
from carrying out their plans for a thoroughgoing modernisation . . . there was
neither the time nor the money to modernise the basic processes or indeed to
adapt the company to the new circumstances. The founders’ faith in the con-
tinuing vitality of the Leblanc process was not borne out by events. The merger
merely slowed down the decay of the industry. This in turn hampered all plans
for improvements because this obsolescent process would not even yield its users
sufficient profits to enable them to scrap it altogether and make a fresh start.
(Haber, 1958, pp. 184–5)
As with the ammonia-soda process, one may ask why the UAC did not switch to
the new technology. The UAC research laboratory did identify the Castner patents
as the best on offer (there were many competing electrolytic processes on offer in
the 1890s) but abandoned negotiations with Castner after 7–8 months. In retro-
spect this decision can be seen to have been fatal. Backed by finance from Solvay,
the Castner-Kellner patents became the basis of a geographic division of markets
for electrolytic producers. In Britain the ‘Castner-Kellner’ company obtained the
monopoly production rights and the UAC was then locked out of the best elec-
trolytic production method for alkali (although by its own account it did try to
develop a proprietary electrolytic method; UAC, 1907).
The decision not to buy the Castner patent is better evidence of the medio-
crity of the UAC’s management than that provided by the ammonia-soda case.
The ICI historian Reader is damning in his judgement of Hurter, the UAC’s head
of research, even suggesting that he had not wanted electrolysis to work when he
presented the results of his research to the UAC board. In Reader’s account,
Hurter distrusted Castner’ s claims and disliked the price being asked for the
patents (Reader, 1970, p. 118). As Reader comments, the irony was that the UAC
had one of the few central research laboratories in Victorian England and they
were adopting best practice by acting on its advice in this matter. Perhaps because
of this, Lindert and Trace are able to quote Muspratt, one of the UAC directors,
dismissing electrolysis as a ‘fad’ in this period (Lindert and Trace, 1971).
Yet Reader acknowledges that years of development of the electrolytic method
were necessary, and somewhat paradoxically given his criticisms of Hurter, that
nrsroxsr +o +rcnxorooic\r coxrr+i+iox 901
© Blackwell Publishers Ltd 2002
‘there was no certainty that Hurter had rejected the final winner’ in 1894 (Reader,
1970, p. 119). Even here we cannot be entirely sure how obvious it could
have been in 1894 that electrolysis was the future for all chlorine and caustic
production and that the Castner patent was clearly the best electrolytic method
at the price Castner offered, or that the UAC’s own efforts to develop electrolysis
would fail.
It is true that UAC’s decision turned out to be a dreadful mistake. Once it was
made, the company became locked into the Leblanc system of production (it
would be more accurate to say, locked out of the substituting technologies) and
became a hapless victim of renewed creative destruction. When Reader describes
UAC as ‘stubbornly attached’ to Leblanc, it is unfair – by the time it was obvious
that Leblanc was doomed, the strategic mistakes had been made and it was too
late for UAC to act.
Conclusion
In an industry like the chemical industry where patents provide effective control
of a technology, the ‘switch’ strategy can be effectively denied to those without
the patents. If ‘in theory’ that leaves the sailing ship effect as a ‘strategy’, we have
found no decisive evidence of it – rather, the major technological change to the
Leblanc system (Chance-Claus process) occurred before the apparently impressive
creation of the UAC and its research laboratory. Despite a vague claim that this
laboratory was able to improve the Leblanc system, no specific innovation could
be cited. Nor is there evidence that the advent of electrolysis ‘induced’ anything
but a rapid closure of Leblanc activity, as resources quickly ebbed away.
A secondary conclusion is that neither the UAC nor the Brunner Mond strat-
egy was evidently ‘irrational’ at any point, as some of the secondary sources con-
clude (Lindert and Trace, 1971; Rothwell and Zegveld, 1985). The creation of
UAC and the agreement reached with Brunner Mond appears quite an achieve-
ment, while the decision of whether to buy the Castner patent was intrinsically
difficult given the need to correctly anticipate the future evolution of the new
technology.
Perhaps what strikes a modern reader is the absence of state action to maintain
the ‘market process’. This state inactivity is a far better candidate for ‘blame’ for the
prolonged survival of UAC and the Leblanc process than irrational entrepre-
neurial behaviour. However, even state inactivity proved not to matter greatly, since
the market agreement appears to have had at best only a delaying effect on the
elimination of Leblanc carbonate production, and then the whole Leblanc process
was rapidly overwhelmed by the arrival of electrolysis.
coxcrtsioxs
The Sailing Ship Effect as a Construction of Hindsight
None of the cases we have analysed provided definitive evidence in support of a
successful sailing ship effect. In our cases, the sailing ship effect is a product of
hindsight assumption and superficial case knowledge; only in the detail is it appar-
ent how innovation in the old technology needed no ‘threat’ from the new. From
the above analysis, I suggest that there appear to be two key problems of inter-
pretation that mislead authors to conclude that there has been a sailing ship effect.
902 ¡. novrrrs
© Blackwell Publishers Ltd 2002
The first is the coexistence of apparently substituting technologies through long periods of
time, which seems to suggest induced improvement in the old technology must have
taken place. What is easily missed in hindsight is that neither of the rival technolo-
gies is static in the period of coexistence and that the pattern of substitution for seg-
mented markets can explain ‘coexistence’. If the new technology evolved in steps and
substituted for segmented markets over time, then ‘coexistence’ is not a problem
requiring explanation (there was never ‘coexistence’ of dissimilar technologies in an
identical market). If the substituting technologies are for a homogenous product and
market, as in the alkali case, then coexistence does require explanation, but in this
bizarre case that explanation did not involve the sailing ship effect, but a ‘two-step’
process of creative destruction.
The second problem with case interpretation is the confusion of continuous innova-
tion in the old technology with innovation induced by the threat of substitution. A claim for
the sailing ship effect must distinguish between continuous innovation driven by
normal intra-industry competition, and innovation that results from the threat of
the substituting technology. In other words, the specific motivation for old tech-
nology innovation must be established in order to demonstrate the sailing ship
effect. None of the sources cited for either of the cases discussed here provide evi-
dence of this kind, which must raise serious doubts about the effect’s existence. Of
course, it is likely to be most difficult to establish such motivation in old, nineteenth
century cases and it is strange that most of the cases commonly cited are indeed
nineteenth century ones – but that is a problem for those who would claim the
effect exists.
Once authors have mistakenly accepted the existence of the effect it becomes
easy to imagine it in other cases. Rothwell and Zegveld, for example, having con-
vinced themselves of the existence of the effect in the alkali case, proceeded to
find it in the case of the substitution of semiconductor transistors for the electronic
valve. A careful reading of Braun and McDonald’s seminal analysis of the semi-
conductor transistor case shows no evidence whatsoever of the sailing ship effect;
rather it demonstrates the difficulty of that other, genuine strategic response, of a
switch to the new technology (Braun and McDonald, 1982).
In the introduction there were references to two other cases purported to
demonstrate the effect: the ice and lighting cases. A brief discussion of these cases
follows in terms of the problems of case interpretation identified here.
A Brief Comment on Gas Versus Electric Lighting and Mechanical Versus Harvested
Ice Production
The process of substitution in the ice case is directly comparable to the sailing ship
case. In the nineteenth century, natural ice was cut from northern US lakes for
shipment south and the greater the distance from the frozen lakes, the greater the
cost of transport and the more ice that melted in transit, hence the higher the
price of ice in southern US states. Yet the demand for ice for uses such as storage
of foods was greatest in precisely the hot regions farthest from the sources of the
ice. This ‘market structure’ created the opportunity for the first primitive and in-
efficient mechanical pump ice machines to establish themselves in the south. As
pumps were improved with new refrigerants and better seals, their efficiency
evolved, the price for mechanically-produced ice dropped and they were able to
become established in more northerly markets. The process of substitution was
incremental and slow and we should not be surprised at continuing old technol-
nrsroxsr +o +rcnxorooic\r coxrr+i+iox 903
© Blackwell Publishers Ltd 2002
ogy innovation in such circumstances; the natural ice harvesters still had to
compete with each other, within their more northerly markets, and as so often in
these cases, production of ice from old and new technology grew together for many
decades.
Utterback cites as examples of old technology innovation during the period of
substitution: the standardization of ice block sizes to facilitate storage and trans-
port, and gasoline-powered circular saws. The first was introduced in the early
nineteenth century and helped consolidate a particular company’s control over
the natural ice trade (Cummings, 1949), but was clearly an innovation related to
the more effective exploitation of natural ice rather than any (then non-existent)
technological threat. The second was introduced in the early 1900s, and
Cummings (one of the historical sources used by Utterback) does state that these
were introduced as a response to mechanically-produced ice (Cummings, 1949,
p. 97). However, Cummings makes this claim in one line of his text; no dates of
adoption are given, no diffusion figures, no justification for the claim that tech-
nological threat was the motivation is given. It is significant that he writes that the
circular saw was ‘made practicable’ by gasoline power – a steam-powered circu-
lar saw had proven uneconomic in the early nineteenth century (Cummings, 1949,
p. 22). Gasoline-powered internal combustion engines had only recently become
available and it was their adaptation to ice-cutting circular saws that enabled this
cost-saving innovation for natural ice harvesting. It is possible that the adaptation
was driven by the substitution threat, but given that an attempt had been made to
introduce the circular saw technology when such a threat did not exist, it is also
possible that the gasoline-powered circular saw was developed and diffused for
normal competitive reasons. Another author, Anderson, acknowledges that these
saws were being introduced in the ‘ordinary course of events’, but claims (without
further evidence) that competition ‘undoubtedly’ accelerated the diffusion
(Anderson, 1953, p. 107). A little more detail is available in Jones, who writes of
a firm, Gifford-Wood, that introduced power saws around 1918, well into the
decline of the natural harvesting industry ( Jones, 1984, p. 65). The firm claimed
its power saws replaced five horse ploughs. Once again, we do not know the think-
ing behind the introduction of this innovation. We know that internal combustion
engines were displacing animals and steam engines as sources of power through-
out the economy in this period of the twentieth century. We know that whatever
the productivity advantages of Gifford Wood’s machine, it did not prevent or
perhaps even delay the eclipse of the harvesting industry – the 1920s saw the bank-
ruptcy of many firms. The ice harvesting industry is hardly a convincing example
of the sailing ship effect.
The gas lighting industry at first sight appears to provide a stunning example
of the sailing ship effect – the radical innovation of the ‘Welsbach mantle’, which
increased gas luminescence by a factor of five, per unit of gas burnt (Bright, 1947).
However, while the Welsbach mantle certainly aided the gas lighting industry, it is
not an example of the sailing ship effect. Welsbach was an independent inventor-
entrepreneur, an Austrian who founded and ran his own electric lighting company
– in Vienna. Nowhere does the historian Bright claim that he developed his mantle
in the USA or with gas company funds or in ‘reaction’ to the threat from electric
lighting.
This story has a further bizarre twist. The thorium and cerium oxides whose
superior fluorescence was responsible for the gas mantle’s luminsescent efficiency
904 ¡. novrrrs
© Blackwell Publishers Ltd 2002
could only be obtained by extraction from ores rich in other rare earth elements
such as tungsten, iridium and osmium. As the gas mantle became standard, and
the market for thorium and cerium grew, the other rare earth metals became cheap
and available for experimentation. So we have the paradox that the success of the
Welsbach gas mantle helped make research possible into rare earth metal fila-
ments, research that would eventually result in the elimination of gas lighting
by the tungsten filament (Bright, 1949, p. 168). Far from Welsbach having any
commitment to the ‘gas industry’, Welsbach himself would contribute to its final
destruction as a source of lighting. He developed his work on the gas mantle by
attempting to pass current through filaments of platinum coated with thorium
oxide. When this failed (because of the different coefficients of thermal expansion)
he produced the first successful metallic filament lamp, which used an osmium
filament (Bright, 1949, p. 174). Millions were sold; it was clearly superior to the
carbon filament lamp that Edison had successfully innovated, and a new general
‘trajectory’ of research was opened.
Welsbach may have helped to prolong the gas industry’s hold over the lighting
market; he then helped to destroy it. He is perhaps better characterized as a leading
‘lighting’ innovator, without particular commitment to the vested gas or electric
lighting technologies. His ability to produce the Welsbach mantle reveals the inad-
equacy of the New York gas companies’ response to Edison’s carbon-filament elec-
tric light, they did not develop the most effective counter-strategy – appropriate
research and development.
The Prospects for the Sailing Ship Effect
This article has necessarily been concerned with just a few cases that purported
to demonstrate the sailing ship effect and has shown that there is no strong empiri-
cal evidence for the effect’s existence and that there are good general grounds for
doubting its widespread existence.
However, all of the cases considered here are examples of the response to radical
technological substitution threats, in nineteenth century ‘pre-modern’ industries
characterized by small, fragmented firms – these appear to have been interpreted
as demonstrations of the sailing ship effect in preference to cases of more modern
technologies. This is a highly selective group of examples; we might seek cases
which varied by having some or all of the following: less ‘radical’ substitution
threats, oligopolistic or monopoly control of markets, multi-technology firms or
firms that had established R&D departments. This is largely beyond the scope
of this paper, but we might have more reason to find the effect in modern
industries.
The best prospect for finding an ‘induced’ acceleration in successful innovation
must be where the actual rate of innovation has fallen below the potential rate for
some reason. Two scenarios where accelerated innovation may be demonstrable
are, firstly, where a fragmented industry structure has prevented sufficient market
control for the foundation of R&D activity. Increased threat may trigger first
cartelization, then the consequent foundation of R&D activity, then an increased
rate of innovation; the first two events actually happened in the alkali case, but we
had no evidence of radical innovation projects intended to improve Leblanc soda
production, and time and resource soon ran out. A second scenario is where there
is monopoly control of the old technology and a stock of unexploited inventions
or foregone but feasible R&D projects exist; in other words, there is ‘technology
nrsroxsr +o +rcnxorooic\r coxrr+i+iox 905
© Blackwell Publishers Ltd 2002
suppression’ (Dunford, 1987). If Dunford’s many examples are valid, these ‘sup-
pressed’ technologies may be more vigorously exploited only when an alternative,
threatening technology emerges that is uncontrolled by the monopoly firm, when
a new entrant to these suppressed technologies appears, or when the state sues the
monopolist.
Finally, if the sailing ship effect is likely to be a rarity in the wider economy this
implies a simplification of the strategic choices of firms facing a substitution threat.
They must decide whether or not to exit the established technology, and if so
whether or not to ‘switch’ to the threatening technology.
nrrrnrxcrs
Axnrnsox, O. E. (1953). Refrigeration in America. London: Oxford University Press.
Bn\tx, E. and M\cnox\rn, S. (1982). Revolution in Miniature. Cambridge: Cambridge
University Press.
Bnion+, A. (1949). The Electric Lamp Industry. New York: Macmillan.
Coorrn, A. C. and Scnrxnrr, D. (1988). ‘Strategic responses to technological threats’. In
Tushman, M. L. and Moore, W. L. (Eds), Readings in the Management of Innovation, 2nd
edition. London: Harper Collins.
Coorrn, A. C. and Sxi+n, C. G. (1992). ‘How established firms respond to threatening
technologies’. Academy of Management Executive, 6, 2.
Ctxxixos, R. O. (1949). The American Ice Harvesters: A Historical Study in Technology,
1800–1918. Berkeley, CA: University of California Press.
Dtxronn, R. (1987). ‘The suppression of technology as a strategy for controlling resource
dependence’. Administrative Science Quarterly, 32, 512–25.
Frr++xrn, A. (1926). The Story of the Rotor. Mt Vernon, NY: FO Willholft.
Fos+rn, R. N. (1988). ‘Timing technological transitions’. In Tushman, M. L. and Moore,
W. L. (Eds), Readings in the Management of Innovation, 2nd edition. London: Harper Collins.
Girrirr\x, S. C. (1935a). Inventing the Ship. Chicago: Follett.
Girrirr\x, S. C. (1935b). The Sociology of Invention. Cambridge, MA: MIT Press.
H\nrn, L. F. (1958). The Chemical Industry during the Nineteenth Century. Oxford: Oxford
University Press.
H\nrrv, C. K. (1971). ‘The shift from sailing ships to steamships, 1850–1890: a study in
technological change and its diffusion’. In McCloskey, D. N. (Ed.), Essays on a Mature
Economy: Britain after 1840. London: Methuen.
Joxrs, J. C. (1984) America’s Icemen – An Illustrative History of the United States Natural Ice
Industry 1665–1925. Humble, TX: Jobeco Books.
Lixnrn+, P. H. and Tn\cr, K. (1971). ‘Yardsticks for Victorian entrepreneurs’. In
McCloskey, D. N. (Ed.), Essays on a Mature Economy: Britain after 1840. London: Methuen.
Rr\nrn, W. J. (1970). Imperial Chemical Industries – A History. London: Oxford University
Press.
Ror+, L. T. C. (1957). Isambard Kingdom Brunel. Harmondsworth: Penguin.
Rosrxnrno, N. (1976). Perspectives on Technology. Cambridge: Cambridge University Press.
Ro+nvrrr, R. and Zro\rrn, W. (1985). Reindustrialisation and Technology. London: Longman.
Scntxrr+rn, J. (1943). Capitalism, Socialism and Democracy. London: Allen and Unwin.
UAC (United Alkali Company) (1907). The Struggle for Supremacy. Liverpool, Walmsley:
United Alkali Company.
U++rnn\ck, J. M. (1996). Mastering the Dynamics of Innovation. Boston, MA: Harvard
Business School Press.
Yt, H. Y. (1982). The Dynamics of Technological Leadership and Economic
Development. M Phil Thesis, Science Policy Research Unit, University of Sussex.
906 ¡. novrrrs
© Blackwell Publishers Ltd 2002

888

. 

from future innovation and elects to exit the threatened market early and to its advantage over ‘forced’ exit. The decision to ‘switch’ from the old to the new technology is particularly interesting and has been the focus for the papers cited above, especially Cooper and Smith (1992). This latter paper examines eight product lines that experienced substitution effects; these range from ball point versus ink pens, to diesel-electric versus steam locomotives. Much of the analysis concerns the behaviour of 27 established firms, selected by Cooper and Smith for their dominant market position in the old technology. All of these entered the new technology, but few managed to establish as dominant a position in the new technology as they had in the old. A diverse range of problems faced those wishing to switch; these included the problems of internal groups which recognised that the advancement of the new technology threatened their expertise and power, and the problems of judging how the new technology would develop and which old competencies could be retained and which should be shed (Cooper and Smith, 1992). Quality of Evidence for the Sailing Ship Effect Cooper and Smith provide the most thorough exploration of response strategies published in the management literature (they consulted over 250 secondary sources). However, they nowhere comment on the third response strategy, the socalled ‘sailing ship effect’, which can be defined as an acceleration of innovation in the old technology in response to the threat of innovation in the new technology. Some earlier management articles on response strategies do refer to the sailing ship effect, but they provide little hard evidence of its occurrence; it seems to be assumed, while the major focus is once again on the strategy of switch. For example, Foster refers to it as a ‘typical response’ (Foster, 1988, p. 220) and describes how sailing ship producers actively sought to improve sail in response to the threat of steam. However, Foster does not give any sources for his assertion and it is likely that he is drawing on the widespread, but incorrect interpretation of Gilfillan, to be discussed later in this paper. A paper by Cooper and Schendel reviewed seven examples of response to technological substitution threat and claimed that firms fought back in two ways: by improving the old technology and by switching to the new. To ‘fight back’ by improving the old technology is a description of the sailing ship effect, but their paper is almost entirely a discussion of the switch decision; they provide no evidence for, or against the existence of the sailing ship effect (Cooper and Schendel, 1988, p. 249). The absence of any reference to the effect in the later Cooper and Smith article (Cooper and Smith, 1992) is evidence of a kind for the absence, rarity, or at least of the effect’s relative unimportance compared to the switch and exit response outcomes. Claims for the effect rely on case study evidence. There is of course the case of steam versus sail in the nineteenth century which gave the effect its name through Gilfillan’s work (Gilfillan, 1935a) and which is referred to by Foster, Rosenberg, and Rothwell and Zegveld. There is also a very strong claim for the existence of the effect in the nineteenth century alkali industry by Rothwell and Zegveld. The alkali industry case also demonstrates the so-called ‘sailing ship effect’ . . . it illustrates how established companies can become locked into existing tech© Blackwell Publishers Ltd 2002

If any historical trend is to be expected. with the Welsbach mantle . alkali. lighting and ice cases. Most fight back. and mechanical versus harvested ice. 1996. p. they are not obviously and systematically different from the other cases. . Although the Cooper and Smith cases are more recent. 1949. such asymmetries may be an important key to a better understanding of the workings of the competitive © Blackwell Publishers Ltd 2002 . Indeed. we have an extraordinary case of economically ‘irrational’ behaviour. but that examples of it were not reported in Cooper and Smith’s review article. A recent book by Utterback devotes a whole chapter to the invasion of stable businesses by radical innovation and concludes that the sailing ship effect is a general strategic response to threatening innovation (Utterback. 1985. Purveyors of established technologies often respond to an invasion of their product market with redoubled creative effort that may lead to substantial product improvement based on the same product architecture. it is surely that firms would have greater ability to respond to invading innovation.    889 nological trajectories. 41) Rothwell and Zegveld not only assert the existence of the effect. . 1996. Of course the established players do not always sit back and watch their markets disappear. it improved gas lighting efficiency by five times and for a time it was not clear that electric lighting would prove the more effective and efficient lighting source (Bright. The imminent threat to a firm’s profit margins which are presented by the rise of a new competing technology seems often in history to have served as a more effective agent in generating improvements in efficiency than the more diffuse pressures of intra-industry competition. p. . The gas companies came back against the Edison lamp . (Rothwell and Zegveld. The gas industry is a particularly promising candidate because the Welsbach gas mantle was certainly a radical innovation. but the economic historian Rosenberg has best articulated an appreciation of the economic significance of the effect if this is so. chapter 7). 126). Why the Sailing Ship Effect is Significant So far a number of authors have claimed that the effect is widespread. p. Rather than attempting to capitalise on the possibilities offered by the emergence of a superior new substitute technology. This case promises to develop our understanding of when and why firms adopt the ‘sailing ship’ strategy. but also that the strategy of accelerating old technology innovation was adopted rather than the superior strategy of ‘switch’ to a demonstrably superior technology. (Utterback. and that response might be expected to include the sailing ship effect. 159) Utterback’s view depends on two more nineteenth century cases: gas versus electric lighting. However. they vigorously defend their position through the accelerated improvement of the old technology. there was nothing incremental about that. it does appear odd that the effect can occur in the sail. If the superiority of the new technology really was known to the old technology firms. .

because we do not know what the sailing ship of the 1880s would have been like in the absence of such inter-technological competition. or search or work differently. but their discussion can be postponed until the widespread nature of the sailing ship effect is established beyond doubt. The sailing ship of the 1880s was far superior to its predecessor of 1850 or so. If these cannot be supported. 1976. (Rosenberg. yet he seems to prevaricate over the sailing ship effect’s existence in the exemplifying case of sail. 205) Although Rosenberg is discussing operational efficiency rather than induced technological change. which is exactly where you would expect to find further evidence. But it seems like a reasonable conjecture. then perhaps either the suppression of innovation is widespread. p. p. We have several strong claims for its existence in four specified cases.890 . clearly if the sailing ship effect is widespread. also seems somewhat dependent on the sail versus steam case. and a peculiar absence of mention of the effect in Cooper and Smith’s thorough review of eight substitution cases. He writes that the builders of sailing ships ‘responded to the competition of iron and steam’ (Rosenberg. Structure and Method of this Paper The approach taken in this paper is to reinvestigate the basis for the strong claims for the effect’s existence in the sail and alkali cases. If the effect is widespread. Obviously one cannot assert this with authority. or perhaps firms are just not sufficiently strongly motivated by normal competition to innovate until threatened with extinction? Such possibilities do not fit well with available theory. 205). Rosenberg. like others. such that they were able to develop genuine radical innovation as a response to threat. © Blackwell Publishers Ltd 2002 .  process. However. 1976. 206) If the effect exists and is widespread. Then in a later passage he becomes more careful about the source of continued sailing ship innovation. Or that when they realized the extremity of the threat to their existence they were somehow so strongly motivated that they were able to organize differently. p. then it seems that it is significant for our understanding of both the general firm strategic response to the threat of substituting radical innovation and the process of competition in the wider economy. and it seems plausible to attribute this improvement to the strong competition of steam. We may conclude at this point that the sailing ship effect is surely worthy of further investigation. 1976. even though they find no place at present in formal economic theory. We have Rosenberg’s observation that operational efficiencies as a result of substitution threats are widespread. it threatens to force us to rethink the very nature of ‘competition’ in the economy. (Rosenberg. is why firms that accelerate technological change as a response to substitution competition did not do so before that competition arose – why did not the normal competitive process between the old technology firms ensure the adoption of any possible innovations in that technology? We are forced to radical alternatives such as that they may have deliberately and collectively suppressed potential innovation until the advent of a major threat. with the sail versus steam case most often cited. One of the strange questions that would be raised if this were so. In the management literature there are no additional developed cases of the effect.

1935b. the metal sail with two stream-lined surfaces. The novel Solvay chemical process technology was from its inception recognized as a radical and superior process to the old Leblanc method of making sodium carbonate. The ‘sailing ship effect’ is not one of Gilfillan’s key principles. in these two cases there will also be a concern to establish how the substitution process could have been interpreted by the authors concerned as evidence of the sailing ship effect. Finally. 1935b. faced by the imminent extinction of all wind-blown vessels. The nearest he comes is in the following passage when he discusses the last shippers to build sailing ships in Europe. there is some discussion of what this implies for our understanding of the concept of technological competition. (Gilfillan. but relevant passages appear when he describes the prospects for sail in the 1930s.    891 and if more plausible and conventional explanations for the appearance of the effect can be found. 1935b). and the rotorship. and Rosenberg cite Gilfillan’s study of innovation in sailing ships as the key example of the sailing ship effect. the sailing barge. 1935a) from which he tries to draw general principles governing innovation in his ‘Sociology of Invention’ (Gilfillan. . sail and gas cases all have invading technologies which evolved through many incremental innovation steps.          Although both Rothwell and Zegveld. Only today. The sail case has been chosen because it is so frequently cited that it has given its name to the effect and it is also an example of a ‘typical’ nineteenth century pattern of incremental and long term technological substitution: the ice. The alkali case has been chosen because it demonstrates a quite different pattern of substitution. This will be followed by an application of the lessons from the sail and alkali cases to the ice and gas cases. . © Blackwell Publishers Ltd 2002 . Only the rate of build of new plant obviously limited the rate of substitution – in this case above all others. p. 18) Today the very threat of extinction seems to have stimulated greater inventiveness than ever . What now follows is a detailed analysis of the competitive process in first the sail and then the alkali case. 163) But Gilfillan nowhere stipulates the motivation or the contribution of the sailing ship producers to the increased effort at innovation. Gilfillan’s work consists of a survey of innovation in ships from earliest history (Gilfillan. 163). (Gilfillan. in Hamburg in the late 1920s (Gilfillan. It will become evident that neither the sail nor alkali cases demonstrate convincing evidence of the sailing ship effect as an acceleration of innovation in the old technology in response to the threat of the new. and the rate of substitution was limited by this slow rate of improvement of the new technology. and especially since the War. p. 1935b. the Leblanc producers must have been aware of the threat of extinction and should have had time to respond. Therefore. p. Gilfillan’s work does not provide an unequivocal demonstration of the effect. do we see signs of an impending revolutionary period – such signs as the auxiliary engine sailer. we will not have to revise our fundamental concepts of competition and strategy.

then steel. Gilfillan and Flettners’ enthusiasm for the renewed future of sail depended largely on the prospects for Flettner’s ‘rotor’ ship. The wind propulsion of ships attracted our interest more than anything else. had produced a feeling of crisis. The established German inventor-entrepreneur Anton Flettner cooperated in creating and then managing the institute. but Gilfillan provides scattered evidence of state intervention and support rather than private sector investment driven by a sense of crisis. . only a few years after the foundation of the institute in 1922 (Gilfillan. More than 60 years later. it tells us nothing about the sailing ship effect. However. Does this mean the sailing ship effect is an imaginative fiction? To complete the analysis we must look at the pattern of real incremental innovations that changed the sailing ship during the period of its displacement by steam. This allowed progressively larger and therefore more economical hulls to be built for both steam and sailing ships. Which is not to say that they did not occur. he provides persuasive evidence that government was the more important funding agency for these attempts at radical invention. was not a fact which one had to accept like a law of nature. not by a sailing ship firm. cited in Gilfillan. © Blackwell Publishers Ltd 2002 . 1926. but neither here nor elsewhere are specific innovations causally associated with their sense of crisis. since we do not know the contribution of private sector. . (Gilfillan. old technology firms to the research. of necessity for action.  The decline of the sailing ship . So Gilfillan associates the sense of crisis within sail with the founding of the Institute voor Aero-en-Hydro-Dynamik in Amsterdam. (Flettner. but by the German Navy. 1935b. p. p. 1935b. but also by the steamship people who valued the windjammers as a training school . p.892 . a research institute with the object of studying the physical flow laws and their industrial application. What is important to our discussion is that the second and most important prototype of the rotor ship was paid for. It will become clear that it is far from clear that such innovation was ‘driven’ by competition from steam. among not only those directly concerned. . even the optimistic Gilfillan comments that the rotor ship appeared to have run into economic difficulties. we know that neither the failed rotor ship nor any of the other projected radical innovations saved the sailing ship in its core markets of freight transportation. p. The Record of Innovation in Sail Much of the clearly important innovation in both steam and sailing ships was derived from the use of cheaper iron. 1935b. 1935b. since in 1926. There was no doubt in our minds that the visibly progressing decay of the sailing ship. Not only does Gilfillan not make an explicit claim for the ‘sailing ship effect’ as a private sector phenomenon. . It cannot have been very much. which could not appreciably be delayed by the advent of the auxiliary (motored) sailing ship. 2l2) Sailing ship producers are included among those who feel the industry has come to a terminal crisis. 212) ‘building had practically ceased’ (Gilfillan. Gilfillan quotes Flettner. By 1935. 163). 2l2) This is a clear expression of the belief that accelerated innovation in sail was possible.

160). Their response was natural. However. This effect was weakened as first coaling stations were established on major trade routes and then eventually new sources of coal developed outside England and Europe. so that although sailing ships declined as a proportion of ships built (in the late 1860s two thirds of tonnage built was sail. This source of economic advantage explains the continually decreasing costs of sail freight contracts recorded at the Baltic exchange (Harley. 1935b. by 1872 it was only 15 per cent. It is clear enough that there is no need to postulate an increased sense of threat to explain continuing innovation of this sort. 1971. the largest tonnage of sailing ships launched in a single year in Britain was launched in 1892 (Harley. 1971).    893 In the case of sailing ships. later. Harley. p. the best we can do is to reconstruct the steps in the pattern of substitution. first. by abolishing the lightest sails. Therefore the pattern of substitution was for first short. cargoes consisting of perishable and high value goods like tea switched to steamers first. Stronger ships and steel masts allowed greater areas of sail and greater wind power to be captured. 1971). wooden hulls were built around a steel frame. and for the shipping of bulk cargoes. aided by the ‘economics’ of trade winds. cutting their crews. Early steam ships were disadvantaged because England was the primary coal source during the nineteenth century and they had to carry their own fuel. In the search for understanding of the competition between the two motive technologies and in the absence of direct evidence of management intentions linked to specific sail innovations. and cutting in two the largest sails until furling © Blackwell Publishers Ltd 2002 . For the same reason. 224). 1971). Gilfillan describes the design of sailing ships adapted to this market niche. This pattern was driven throughout the period of coexistence of the two motive technologies by constant improvements to the efficiency of marine engines and boilers (boiler operating pressure was crucial and depended on techniques of metal manufacture). What we have called the ‘period of substitution’ lasted for over 70 years and in this time the volume of shipping grew enormously. The market for shipping services also became increasingly segmented over this period. Relative decline was masked by absolute growth in demand for many decades and this effect. the more cargo space had to be allocated to coal storage. . common to many substitution cases. of low tonne value (Harley. then increasingly long distance voyages from England to be taken by steam ships. which is why passenger services were taken by steam as early as the 1850s and 1860s (Harley. 1971) and somewhat obscures the real substitution process of steam as a motive force for sail. steam cargo rates rose dramatically with distance from England – the greater the journey from England. . As a consequence. broadening the upper yard . where firms with similar technologies seek to better meet the needs of users through the improvement of their products and services – in this case through the exploitation of cheapening new materials. the usual competitive process is sufficient. The most obvious feature of this process was that substitution was hampered by the slow evolution of steam as a power source. hulls and sail masts could be made entirely of steel (Gilfillan. might be expected to mask the ‘threat’ from steam. like coal. sail remained economic compared to steam into the twentieth century for the longest voyages to the pacific coast of the USA and to Australia. The principal service advantage of steam over sail lay in its speed and reliability. p.

our problem is that these operational adaptations were not necessarily made ‘in competition’ with steam. As far as the sailing ship effect is concerned. perhaps. small crews became possible. i. The only examples we have of such statements are associated with. In order to think about this. (Gilfillan. for it may just as well have been driven by ‘normal’ competition between old technology incumbents. The long term and generally incremental nature of the substitution process and the rise in total market size for both sail and steam until the mid-1890s suggest that for much of this period the ‘perceived substitution threat’ on the part of sailing ship producers would have been relatively weak. p. in deep seas and with unchanging. might produce this kind of evidence. They might just as well have been adaptations to the physical characteristics of long distance voyages. 160) Other adaptations were made. The Lack of Evidence for the Effect Throughout this detailed account of the substitution process. all we have is Harley’s chart of a monotonic decline in sailing ship freight costs. which provided reliable motive force. It is Unlikely that a Strong Threat was Perceived Throughout the Period of Substitution The plausibility of the sailing ship effect hinges first on the perception on the part of sailing ship producers that there was a strong threat to their existence. On the very long voyages which comprised the remaining ‘market’ for sailing ships. driven by competition between rival sailing ship firms for contracts. the last established type. but it remains possible that a thorough search of specialized library archives in. The © Blackwell Publishers Ltd 2002 . and more use made of the easily handled fore-and-aft sails. the Flettner rotor ship effort. which if it is anything is evidence against the sailing ship effect. Decisive evidence might exist in primary company sources and would consist of management statements that innovative projects were successfully carried through as a response to steam. such as auxiliary steam engines to further reduce labour costs and to increase manoeuvrability. became almost standard around 1900. the listed innovations aided the efficient exploitation of trade winds. Given this situation it is still possible to discuss the characteristics of the substitution process as they would be likely to impact on the sailing ship effect ‘decision’. 1935b. The evidence of clear radical invention like the rotor ship (which failed) was financed and driven by government. as a conscious strategic response to declining costs of steam transport. for example in harbour. There is no ‘innovation frequency chart’ to demonstrate an increase in the rate of innovation in response to steam. but alternatives to each other only at some geographical ‘boundary’ between these markets. we have this problem that continuing old technology innovation is not evidence of the sailing ship effect. not the firms supposedly experiencing a great threat.e. Hamburg. Over a period of more than half a century this boundary moved to eliminate the sailing ship – but at any point in time the bulk of orders for sail and steam ships would be for a market that was not imminently threatened by substitution. once again. a government-inspired project.894 .  could replace almost all reefing. constant winds. so that the great steel four-mast bark. it may be useful to think of the two technologies as dominant in two separate markets. Sails and mast heights were standardised. like the noble Brilliant.

We would like to know how difficult it was to switch between the manufacture of sailing ships and steam ships. Unlike sail. alkali was an undifferentiated commodity product sold into a national market. In conjunction with the observation that ‘normal’ competition within the old technology should result in adoption of old technology innovations. 1935b) and by this time it was apparently too late for producers to act. the case of rival alkali production technologies in late nineteenth century England. We also recall that by the early twentieth century. It is Likely that the Strategic Alternatives of ‘Switch’ and ‘Exit’ Would Be Compared to the (Possible) Option of the Sailing Ship Effect One last issue pertinent to the sailing ship effect may be raised in this case. sailing ship hulls were made from steel as steamships were – sailing ship producers were acquiring some of the technological abilities that steamship builders would need. If it were easy to switch from the old to the new technology. In general. and we may therefore doubt that when it was finally recognized that sailing ships were on the point of elimination there would have been a strong desire to react in their defence (whether or not possible). or the degree of technological ‘lock-in’. We turn now to the other case that has been interpreted as providing strong evidence of the effect. even if possible. 1957). Rolt. Brunel’s Great Western and Eastern steamships. would – or should – be compared to the alternative strategies of ‘switch’ to the new technology – and of course. we might expect that the effort of the sailing ship effect. Here we have a case where a strong sense of threat was experienced by the old technology firms and yet the strategy of ‘switch’ was blocked by the patent position of the firm controlling the new technology. which implies that the potential for increasing the rate of innovation in the old technology is low to non-existent. one would expect that the competition between the two processes should be especially intense and resolved relatively quickly – only limited by the time taken © Blackwell Publishers Ltd 2002 . If the Solvay process was clearly superior to the Leblanc. if it exists at all. the rival processes for producing alkali (sodium carbonate) coexisted for decades. the ‘strategy’ of exit to an entirely different activity. All this may be taken to suggest that the degree of lock-in to the old technology was certainly not insurmountable.                As in the sail case. in the mid-nineteenth century the same producers also equipped early (and unreliable) steamships with sail (for example. it would be a matter of relative indifference to producers if the old technology were eliminated – they would simply move to the new technology.    895 only evidence we have that producers experienced a strong sense of threat is at the very end of the substitution period (Gilfillan. We have a strong case for the absence of the sailing ship effect in the case that above all others was thought to exemplify the effect. we can suspect that the sailing ship effect. We do have some piecemeal evidence that the sailing ship producers were not strongly locked-in to the old technology. is not widespread. the strength of the sailing ship builders’ commitment to the old technology.

Leblanc soda production began a rapid decrease. However. As with sail and steam. below. were unwilling to abandon it [ Yu’s emphasis]. despite the rapidly growing market for soda ash throughout the second half of the nineteenth century. Lindert and Trace (although they have nothing to say about the sailing ship effect.  to build new Solvay plant – or unless there was a sailing ship effect. 1971). 1985. cited in Rothwell and Zegveld. for example. but only poor judgement. Figures compiled (Table I) from a standard chemical industry history (Haber. as does a third source. Like Yu. by concentration through the merging of companies to form the United Alkali Company. the superior and soon dominant ammonia-soda method of sodium carbonate manufacture. Production began in halting fashion in 1874 (Reader. it will be shown that an alternative and better explanation is available for the adherence to the old technology that involves no irrationality and no sailing ship effect. This was clearly in the Leblanc producers interests – it helped prolong the life of the Leblanc system and is a partial explanation of the prolonged period of coexist© Blackwell Publishers Ltd 2002 . Instead. 53). and through entering into price fixing agreements with British Solvay producers ( Yu. p. 1982. The conditions of the patent were such as to make Brunner Mond the effective monopoly supplier of Solvay ammonia-soda in Britain. This begs for an explanation and with no further information there is a tendency among some of the secondary sources to assume a successful (in the sense of some economic gain) sailing ship strategy on the part of the Leblanc producers. Brunner Mond had entered a market fixing agreement with the Leblanc producers despite the very evident success of the Solvay process against the old technology. a more detailed analysis of the case is required. there was an apparent protracted 42 year period of coexistence of the two technologies. who had a large investment in Leblanc plant. Yu. Since Brunner Mond obtained a licence for the Solvay process and began production in Britain in 1872. British managers. thus facilitating economies of scale. Instead there is a dramatic cessation of expansion of Solvay production by Brunner Mond around 1890 and static output for eight years. p. Rothwell and Zegveld also stress the apparent irrationality of the Leblanc producers. and crude extrapolation of trends in Table I suggest that it would have been possible for creative destruction to have been completed before 1900. Lindert and Trace. 41).896 . 1958) show creative destruction well underway in the 1880s. From Creative Destruction in a Free Market to a Rigged Market Brunner Mond had a preferential license in Britain for the Solvay process. they responded to the challenge in a number of ways: by increasing efficiency through process improvement. 1970. The unexplanned mystery is why Leblanc producers were prepared to do all this when the better available strategy was to switch to Solvay production. 1971). but the last Leblanc plant was only closed in the first world war (Lindert and Trace. Once Brunner Mond production became significant. The long period of coexistence of the two processes when one is obviously superior to the other is a problem requiring explanation and superficially it does appear to support the conclusion of irrationality on the part of the Leblanc producers. Despite the obvious advantages of the Solvay process.

with the proven superior technology. as in the case of sail.1 1886 165. The Leblanc process had more wastes and byproducts than the Solvay process. It will be argued that in the detail of such changes lies the explanation of the Leblanc producers’ apparently ‘irrational’ adherence to the Leblanc process. as has been assumed by some of the secondary sources (in particular. 1971. we need to know more detail of the Leblanc and Solvay processess and the events behind their substitution.9 897 Data derived from Haber (1958. 1958. for which there was a new and growing market. Solvay versus Leblanc production of sodium carbonate in Britain in thousands of tons Solvay (ktons) (Haber. which produced hydrochloric acid. Later. Rothwell and Zegveld). 158) 1878 4 1880 18. 152) 1878 196. should want to enter an agreement with the ‘loser’ firms. ence of the rival technologies. The second stage of the Leblanc process involved the reduction of sodium sulphate with limestone and coal or coke to produce either the carbonate or the hydroxide (caustic soda) and ‘tank waste’. Solvay as a Partial Technological Substitute for Leblanc Output The key to understanding this case is that it is not a simple one of substitution of one process for sodium carbonate production by another. This was originally discarded.5 1898 181 1900 225 1903 240 Leblanc (ktons) (Haber. Both processes produced byproducts and wastes whose economic value changed through this period.5 1890 179. p. 158).8 1885 77. especially as the alkali industry continued to differentiate into caustic.    Table I. while UK government statistics do not distinguish between the output of Solvay and Leblanc processes. pp. 53) and 1900 figure from Reader (1970. This table represents a compilation of all output data which specifies the producing technology that I have been able to find in the sources listed. 282). NB: The UAC refused to publish precise Leblanc soda output figures from 1893 (Lindert and Trace. p. p.9 1880 266. 174). p. So caustic soda could be made at the expense of soda production and was more cheaply © Blackwell Publishers Ltd 2002 . which was mostly calcium sulphide. p. we have to be careful about what ‘competes’ with what. 152. Additional data: 1878 Solvay figure from Reader (1970.2 1884 204. bleaching powder and chlorine products as well as sodium carbonate. but by the second half of the nineteenth century it could be used as a key input for the manufacture of bleaching powder. 1958. Output figures later than this date are difficult to establish. It began with the treatment of salt with sulphuric acid. chlorine could be extracted for another distinct market via the Deacon process. To understand this firm behaviour. it is now not clear why Brunner Mond.1 1882 233. However.

106). 1970. which restored and maintained overall profitability for both processes until at least 1897 (Lindert and Trace. p. p. 1971. Brunner Mond agreed an upper limit to its soda production of 165. but by limiting UAC entry into ammonia-soda production preserved the identity of each of the firms with the major ‘rival’ technologies. Irrespective of the details of the agreement or its effective prosecution. p. alkali prices began a speculative rise. the details of the agreement do reveal the strategic concerns of the two producers and in particular provide some explanation for the otherwise astonishing entry of Brunner Mond into the market agreement when it possessed the superior production technology. p. the Leblanc producers would retain a limited interest in the state of the sodium carbonate market.898 . 1970. 1971. so reducing losses and boosting profits at the same time (Haber. 1970. In return. 108). 1958. Brunner Mond’s great vulnerability was the expiry of its major Solvay patent in 1886 (Reader. whenever bleaching powder was made. either sodium carbonate or caustic soda would have to be made. 1971. p. which meant that © Blackwell Publishers Ltd 2002 .000 tons/year. a scale of decline greater than any estimated cost difference between the two soda production processes. These ‘friendly relations’ comprised an ‘agreement on price levels and market shares for each soda product’ (Lindert and Trace. and while this aggravated losses for the many Leblanc producers. 1958. 1970. 257). 106). It is this collapse of profitability for all producers that induced the restructuring of the alkali industry and the collective rigging of output and prices of the main alkali products. p. to limit production of chlorine and not to enter the market for caustic (Reader. 182). that they ‘made it clear they intended to effect far-reaching reforms’ and that they were motivated to ‘establish friendly relations with the Solvay manufacturers’ (Haber. 1970. p. on circulation of rumours of the formation of the UAC. the Leblanc producer deliberately switched production from soda. p. 54). UAC subsequently made a limited entry into ammonia-soda technology with the purchase of two ammonia-soda plant in 1893 (Reader. The agreement therefore effectively ended creative destruction in the carbonate market. However. 1971).000 tons/year. 251). it also brought near zero profit for Brunner Mond by 1887 (Lindert and Trace. and with the creation of the giant UAC (1891) a full or partial switch from Leblanc into ammonia soda production was at least a theoretical possibility. The Logic of the Market Agreement Prices for soda ash in Europe fell over 75 per cent between the 1860s and late 1880s (Reader. p. the decline accelerated. Brunner Mond had further reason to enter an agreement since Solvay soda was only a partial technological substitute for the Leblanc process. However. The more than 40 independent Leblanc producers reacted to the slump in prices by merging in 1891 into the United Alkali Company (UAC) (Reader. As demand for caustic grew. 109). 253) and by its periodic renegotiation and renewal the two parties maintained the rigged market into the first world war. Haber writes of the promoters of the UAC. 96). p.  made via the Leblanc process than the Solvay (Lindert and Trace. In the 1891 agreement Brunner Mond obtained an agreed upper limit for UAC ammonia-soda production of 15. With a general trade depression between 1884 and 1889. This meant that despite the growth of these new markets.

It depended far more on detailed experience of the process in action and on trade secrets. UAC’s strategy of negotiating a rigged market. which itself had a growing diversity of uses through this period. 1958. depended comparatively little on the kind of knowledge which is revealed in a patent specification or expounded in a textbook. But if it came to a price war. dispersed technical assistance and new patents within the group. These secrets were and are very carefully guarded indeed . .    899 Brunner Mond could not expect to achieve the full creative destruction of the Leblanc producers through output expansion. 100). . . One last deterrent faced a UAC strategy of massive entry into ammonia-soda – the technical difficulty of operating ammonia-soda plant without Solvay group assistance. it was too late for a strategy of wholesale switch into ammonia-soda. Brunner Mond had capital reserves while UAC was loaded with debt (Haber. p. 45) One of the radical features of the Solvay process was that it operated continuously and much of the difficulty of management was connected to this feature. . Nor was an aggressive switch to ammonia-soda an obvious and profitable strategy for the UAC. This could be recycled to the production of sulphuric acid. 1958. We have no clear evidence that the development of the Claus-Chance process was driven by the threat from Solvay production. . it will be evident that the Solvays’ success. By 1891 Brunner Mond held over 50 per cent of British carbonate production (Table I). 1970. The Solvay group collected details on individual plant performance at fortnightly intervals and in return. and before the arrival of © Blackwell Publishers Ltd 2002 . Given the years it had taken the Solvay brothers to build up this knowledge and given Brunner Mond’ s powerful lead in operating the technology. the chemical histories tell us that there had been many attempts over decades. Such a policy risked the worst of all worlds: a low sodium carbonate price and the danger of retaliatory entry into Solvay technology financed by the premium prices obtainable in UAC’s monopoly over the bleaching powder and caustic soda markets. 159). Rather. The Reaction of the Leblanc Producers to Solvay Production – Dubious Evidence of the Sailing Ship Effect The significance of these events for the Leblanc system is that it bought time and increased profits with which the Leblanc producers might have sought to improve their position. maintaining a division of control of the technologies and placing their hopes on the future of the chlorine and caustic markets looks reasonable. Major entry into ammonia-soda risked a further price collapse and war with Brunner Mond. (Reader. combined with what appeared to be time and resource – what did they in fact do? There were improvements to the Leblanc system. but the greatest technological innovation of the period of substitution came before the formation of the UAC and was the Claus-Chance process of 1882. In essence. by the time UAC was created and could pose as a threat to Brunner Mond. They had a strong motivation to act. p. when at length it came. Reader comments on the development of the Solvay process that. p. which made it possible to recover the sulphur previously lost in the form of calcium sulphide (Haber.

There was also some diversification – UAC bought a railway in Spain (Reader. 41). they came to define themselves as primarily caustic and bleaching powder producers. p.  the Solvay process. As in the case of sail. Of this we have no evidence. Nor can the UAC be easily judged as irrational by its decision to restrict itself to the development of these markets – they appeared to be secure from Solvay competition. Instead the efforts at improving operational performance were primarily intended to benefit the development of the bleaching powder. p. the latter to secure sulphur supplies. which if it suggests any strategy. It is now clear that the ‘paradox’ of the long period of coexistence of Solvay and Leblanc production in Britain is no paradox at all. perhaps as early as 1890. 1971. Improvements in the economics of soda production did occur. UAC. they are unlikely to have been of great economic importance. but a chlorine and caustic soda technology. caustic and sulphuric acid markets and appear more as strategies of ‘exit’ rather than the sailing ship effect. p. If they existed. 41). 1907. The problem with the actions listed above is that while some are efforts to improve the Leblanc system of production. Other UAC activities that are described are vertical integration into salt and pyrites mines. If we knew what these were. but they are nowhere specified in published company documents (for example. 1907. So the UAC claimed that its creation subjected the constituent Leblanc manufacturing plant to central comparison and control and this allowed it to establish best operating practice in every works (UAC. The ‘late’ (1890s) Leblanc system was not primarily a sodium carbonate production technology as it had once been. who happened to make some high cost sodium carbonate as a byproduct of these operations. but by their own published comments it appears that by 1902 at the latest the UAC had ceased to make carbonate altogether (Lindert and Trace. 1971. 1907. This does not really count as the sailing ship effect. as a description of subsequent events will show. It is reasonable to assume that some time earlier. 249. This is another example of that ‘tightening’ of operational efficiency that Rosenberg noted as common in substitution cases and that we may have seen in the sail case. 282). We defined the sailing ship effect as if two substitutable products would compete and on this basis. 1907) or the chemical histories. to recover the valuable sulphur lost in the Leblanc process. continuing old technology innovation does not imply the existence of the sailing ship effect. UAC. but they became incidental to the main strategy of exit from soda production – they were never expected to enable serious price ‘competition’ with the Solvay process. the main sources cite UAC’s claims at the time of its formation that it would ‘modernize’ and ‘reform’ its processes. they are not necessarily efforts to improve the economics of sodium carbonate production. p. Texts like Haber comment only that theoretical cost savings and potential new markets always provided the incentives to innovate and improve the Leblanc system. we would have examples of the sailing ship effect. for the sailing ship effect to exist we would require evidence of efforts to accelerate innovation specifically in Leblanc soda production. which is properly the induction of technological change – involving change in the old artefact forms. which at first appears promising. However. 33). p. The UAC did not publish output figures after 1893 (Lindert and Trace. © Blackwell Publishers Ltd 2002 . UAC. 1907). The UAC did found a ‘Central Research Laboratory’. suggests increased effort to exit Leblanc production. 1970.900 . which it claimed produced improvements in old processes (UAC.

one of the UAC directors. This in turn hampered all plans for improvements because this obsolescent process would not even yield its users sufficient profits to enable them to scrap it altogether and make a fresh start. 1958. there was neither the time nor the money to modernise the basic processes or indeed to adapt the company to the new circumstances. 1907). p. In retrospect this decision can be seen to have been fatal. 158). 1971). The merger merely slowed down the decay of the industry. 1958. p. . the Castner-Kellner patents became the basis of a geographic division of markets for electrolytic producers. The founders’ faith in the continuing vitality of the Leblanc process was not borne out by events. As Haber describes the period of precipitous decline. the UAC’s head of research. pp. and somewhat paradoxically given his criticisms of Hurter. the decisive blow was the development of modern electrolytic production methods. Electrolysis of strong brine solution enabled the direct production of caustic soda and chlorine well below Leblanc costs. 184–5) As with the ammonia-soda process. that © Blackwell Publishers Ltd 2002 . The UAC research laboratory did identify the Castner patents as the best on offer (there were many competing electrolytic processes on offer in the 1890s) but abandoned negotiations with Castner after 7–8 months. Hurter distrusted Castner’ s claims and disliked the price being asked for the patents (Reader. The formation of the UAC and the Brunner Mond market sharing agreement proved to provide a very short reprieve for the Leblanc system – so short that it casts further doubt on the UAC’s ability to deliver any real improvements to the Leblanc system. . 1958. In Reader’s account. one may ask why the UAC did not switch to the new technology. The ICI historian Reader is damning in his judgement of Hurter. Perhaps because of this. Although Brunner Mond developed a means of competing in the chlorine and caustic markets and used these to renegotiate the alkali market sharing agreements in its favour (Haber. Yet Reader acknowledges that years of development of the electrolytic method were necessary. Backed by finance from Solvay. it rapidly became too late for the Leblanc producers to act. (Haber. Lindert and Trace are able to quote Muspratt. p. Electrolytic production escalated from 1895 and rapidly eroded the remaining profitable Leblanc markets (Haber. The decision not to buy the Castner patent is better evidence of the mediocrity of the UAC’s management than that provided by the ammonia-soda case. 1970.    901 Electrolytic Production of Alkali – Extinction of Leblanc Unfortunately for the UAC this situation did not last long because new sources of technological competition eroded its monopoly position in chlorine and caustic soda product markets. 160). UAC. In Britain the ‘Castner-Kellner’ company obtained the monopoly production rights and the UAC was then locked out of the best electrolytic production method for alkali (although by its own account it did try to develop a proprietary electrolytic method. Once electrolytic production began to affect market prices. the irony was that the UAC had one of the few central research laboratories in Victorian England and they were adopting best practice by acting on its advice in this matter. As Reader comments. even suggesting that he had not wanted electrolysis to work when he presented the results of his research to the UAC board. dismissing electrolysis as a ‘fad’ in this period (Lindert and Trace. The board soon discovered that the discouraging trading results prevented them from carrying out their plans for a thoroughgoing modernisation . 118).

as resources quickly ebbed away. Conclusion In an industry like the chemical industry where patents provide effective control of a technology. Even here we cannot be entirely sure how obvious it could have been in 1894 that electrolysis was the future for all chlorine and caustic production and that the Castner patent was clearly the best electrolytic method at the price Castner offered. while the decision of whether to buy the Castner patent was intrinsically difficult given the need to correctly anticipate the future evolution of the new technology. I suggest that there appear to be two key problems of interpretation that mislead authors to conclude that there has been a sailing ship effect. Perhaps what strikes a modern reader is the absence of state action to maintain the ‘market process’. 1971. no specific innovation could be cited. only in the detail is it apparent how innovation in the old technology needed no ‘threat’ from the new. Rothwell and Zegveld. and then the whole Leblanc process was rapidly overwhelmed by the arrival of electrolysis.902 . even state inactivity proved not to matter greatly. we have found no decisive evidence of it – rather. locked out of the substituting technologies) and became a hapless victim of renewed creative destruction. In our cases. since the market agreement appears to have had at best only a delaying effect on the elimination of Leblanc carbonate production. If ‘in theory’ that leaves the sailing ship effect as a ‘strategy’. 1970. However. A secondary conclusion is that neither the UAC nor the Brunner Mond strategy was evidently ‘irrational’ at any point. the major technological change to the Leblanc system (Chance-Claus process) occurred before the apparently impressive creation of the UAC and its research laboratory. or that the UAC’s own efforts to develop electrolysis would fail. Nor is there evidence that the advent of electrolysis ‘induced’ anything but a rapid closure of Leblanc activity. the strategic mistakes had been made and it was too late for UAC to act. From the above analysis. the company became locked into the Leblanc system of production (it would be more accurate to say. Once it was made.  ‘there was no certainty that Hurter had rejected the final winner’ in 1894 (Reader. it is unfair – by the time it was obvious that Leblanc was doomed. the sailing ship effect is a product of hindsight assumption and superficial case knowledge. the ‘switch’ strategy can be effectively denied to those without the patents. © Blackwell Publishers Ltd 2002 .  The Sailing Ship Effect as a Construction of Hindsight None of the cases we have analysed provided definitive evidence in support of a successful sailing ship effect. as some of the secondary sources conclude (Lindert and Trace. 1985). It is true that UAC’s decision turned out to be a dreadful mistake. Despite a vague claim that this laboratory was able to improve the Leblanc system. When Reader describes UAC as ‘stubbornly attached’ to Leblanc. p. The creation of UAC and the agreement reached with Brunner Mond appears quite an achievement. This state inactivity is a far better candidate for ‘blame’ for the prolonged survival of UAC and the Leblanc process than irrational entrepreneurial behaviour. 119).

genuine strategic response. which must raise serious doubts about the effect’s existence. of a switch to the new technology (Braun and McDonald. rather it demonstrates the difficulty of that other. as in the alkali case. it is likely to be most difficult to establish such motivation in old. This ‘market structure’ created the opportunity for the first primitive and inefficient mechanical pump ice machines to establish themselves in the south. and innovation that results from the threat of the substituting technology. If the substituting technologies are for a homogenous product and market. Yet the demand for ice for uses such as storage of foods was greatest in precisely the hot regions farthest from the sources of the ice. In the introduction there were references to two other cases purported to demonstrate the effect: the ice and lighting cases. the greater the cost of transport and the more ice that melted in transit. The second problem with case interpretation is the confusion of continuous innovation in the old technology with innovation induced by the threat of substitution. their efficiency evolved. proceeded to find it in the case of the substitution of semiconductor transistors for the electronic valve. then ‘coexistence’ is not a problem requiring explanation (there was never ‘coexistence’ of dissimilar technologies in an identical market). In other words. natural ice was cut from northern US lakes for shipment south and the greater the distance from the frozen lakes. nineteenth century cases and it is strange that most of the cases commonly cited are indeed nineteenth century ones – but that is a problem for those who would claim the effect exists. the price for mechanically-produced ice dropped and they were able to become established in more northerly markets. Rothwell and Zegveld. 1982). A brief discussion of these cases follows in terms of the problems of case interpretation identified here. In the nineteenth century. the specific motivation for old technology innovation must be established in order to demonstrate the sailing ship effect. for example. What is easily missed in hindsight is that neither of the rival technologies is static in the period of coexistence and that the pattern of substitution for segmented markets can explain ‘coexistence’. A careful reading of Braun and McDonald’s seminal analysis of the semiconductor transistor case shows no evidence whatsoever of the sailing ship effect. but a ‘two-step’ process of creative destruction. Of course. As pumps were improved with new refrigerants and better seals. having convinced themselves of the existence of the effect in the alkali case. hence the higher the price of ice in southern US states. Once authors have mistakenly accepted the existence of the effect it becomes easy to imagine it in other cases. A claim for the sailing ship effect must distinguish between continuous innovation driven by normal intra-industry competition. which seems to suggest induced improvement in the old technology must have taken place.    903 The first is the coexistence of apparently substituting technologies through long periods of time. If the new technology evolved in steps and substituted for segmented markets over time. but in this bizarre case that explanation did not involve the sailing ship effect. None of the sources cited for either of the cases discussed here provide evidence of this kind. then coexistence does require explanation. The process of substitution was incremental and slow and we should not be surprised at continuing old technol© Blackwell Publishers Ltd 2002 . A Brief Comment on Gas Versus Electric Lighting and Mechanical Versus Harvested Ice Production The process of substitution in the ice case is directly comparable to the sailing ship case.

p. we do not know the thinking behind the introduction of this innovation. Welsbach was an independent inventorentrepreneur. A little more detail is available in Jones. the natural ice harvesters still had to compete with each other. it is also possible that the gasoline-powered circular saw was developed and diffused for normal competitive reasons.  ogy innovation in such circumstances. acknowledges that these saws were being introduced in the ‘ordinary course of events’. no dates of adoption are given. 1949. no diffusion figures. 1984. 1953. The ice harvesting industry is hardly a convincing example of the sailing ship effect. 1949. It is possible that the adaptation was driven by the substitution threat. Once again. who writes of a firm. while the Welsbach mantle certainly aided the gas lighting industry. no justification for the claim that technological threat was the motivation is given. The first was introduced in the early nineteenth century and helped consolidate a particular company’s control over the natural ice trade (Cummings. an Austrian who founded and ran his own electric lighting company – in Vienna. well into the decline of the natural harvesting industry ( Jones. p. However. We know that internal combustion engines were displacing animals and steam engines as sources of power throughout the economy in this period of the twentieth century. The thorium and cerium oxides whose superior fluorescence was responsible for the gas mantle’s luminsescent efficiency © Blackwell Publishers Ltd 2002 . Another author. We know that whatever the productivity advantages of Gifford Wood’s machine. 65). Utterback cites as examples of old technology innovation during the period of substitution: the standardization of ice block sizes to facilitate storage and transport. 1947). Cummings makes this claim in one line of his text. 22). This story has a further bizarre twist. which increased gas luminescence by a factor of five. and as so often in these cases. but given that an attempt had been made to introduce the circular saw technology when such a threat did not exist. The firm claimed its power saws replaced five horse ploughs. but was clearly an innovation related to the more effective exploitation of natural ice rather than any (then non-existent) technological threat. that introduced power saws around 1918. It is significant that he writes that the circular saw was ‘made practicable’ by gasoline power – a steam-powered circular saw had proven uneconomic in the early nineteenth century (Cummings. p. within their more northerly markets. but claims (without further evidence) that competition ‘undoubtedly’ accelerated the diffusion (Anderson. Gasoline-powered internal combustion engines had only recently become available and it was their adaptation to ice-cutting circular saws that enabled this cost-saving innovation for natural ice harvesting. production of ice from old and new technology grew together for many decades. and gasoline-powered circular saws. per unit of gas burnt (Bright. 1949). Gifford-Wood. 107). it is not an example of the sailing ship effect. Nowhere does the historian Bright claim that he developed his mantle in the USA or with gas company funds or in ‘reaction’ to the threat from electric lighting. However.904 . it did not prevent or perhaps even delay the eclipse of the harvesting industry – the 1920s saw the bankruptcy of many firms. The second was introduced in the early 1900s. The gas lighting industry at first sight appears to provide a stunning example of the sailing ship effect – the radical innovation of the ‘Welsbach mantle’. Anderson. and Cummings (one of the historical sources used by Utterback) does state that these were introduced as a response to mechanically-produced ice (Cummings. 97). p.

This is largely beyond the scope of this paper. Welsbach himself would contribute to its final destruction as a source of lighting. Far from Welsbach having any commitment to the ‘gas industry’. which used an osmium filament (Bright. p. The Prospects for the Sailing Ship Effect This article has necessarily been concerned with just a few cases that purported to demonstrate the sailing ship effect and has shown that there is no strong empirical evidence for the effect’s existence and that there are good general grounds for doubting its widespread existence. fragmented firms – these appear to have been interpreted as demonstrations of the sailing ship effect in preference to cases of more modern technologies. but we might have more reason to find the effect in modern industries. However. iridium and osmium. Two scenarios where accelerated innovation may be demonstrable are. he then helped to destroy it. the first two events actually happened in the alkali case. without particular commitment to the vested gas or electric lighting technologies. As the gas mantle became standard. firstly. and the market for thorium and cerium grew. in nineteenth century ‘pre-modern’ industries characterized by small. He is perhaps better characterized as a leading ‘lighting’ innovator. 174). but we had no evidence of radical innovation projects intended to improve Leblanc soda production. p. they did not develop the most effective counter-strategy – appropriate research and development. Welsbach may have helped to prolong the gas industry’s hold over the lighting market. then the consequent foundation of R&D activity. 1949. A second scenario is where there is monopoly control of the old technology and a stock of unexploited inventions or foregone but feasible R&D projects exist. where a fragmented industry structure has prevented sufficient market control for the foundation of R&D activity. the other rare earth metals became cheap and available for experimentation. When this failed (because of the different coefficients of thermal expansion) he produced the first successful metallic filament lamp. Millions were sold. there is ‘technology © Blackwell Publishers Ltd 2002 . and a new general ‘trajectory’ of research was opened. all of the cases considered here are examples of the response to radical technological substitution threats. then an increased rate of innovation. in other words. 1949. multi-technology firms or firms that had established R&D departments. we might seek cases which varied by having some or all of the following: less ‘radical’ substitution threats. 168). and time and resource soon ran out. His ability to produce the Welsbach mantle reveals the inadequacy of the New York gas companies’ response to Edison’s carbon-filament electric light. He developed his work on the gas mantle by attempting to pass current through filaments of platinum coated with thorium oxide. This is a highly selective group of examples. Increased threat may trigger first cartelization.    905 could only be obtained by extraction from ores rich in other rare earth elements such as tungsten. So we have the paradox that the success of the Welsbach gas mantle helped make research possible into rare earth metal filaments. research that would eventually result in the elimination of gas lighting by the tungsten filament (Bright. oligopolistic or monopoly control of markets. The best prospect for finding an ‘induced’ acceleration in successful innovation must be where the actual rate of innovation has fallen below the potential rate for some reason. it was clearly superior to the carbon filament lamp that Edison had successfully innovated.

1800–1918. Imperial Chemical Industries – A History. A. when a new entrant to these suppressed technologies appears. J. L. (1943). 512–25. London: Allen and Unwin.). W. R. C. Y. Mastering the Dynamics of Innovation. H. TX: Jobeco Books. The Dynamics of Technological Leadership and Economic Development. London: Oxford University Press. (1957). MA: MIT Press. C. (1988). C. and S. B. N. UAC (United Alkali Company) (1907). Refrigeration in America. D. Chicago: Follett. University of Sussex. U. R. Inventing the Ship. L. New York: Macmillan. C. Perspectives on Technology. Humble. Cambridge: Cambridge University Press. T. Essays on a Mature Economy: Britain after 1840. S. L. Cambridge. Y. London: Oxford University Press. R. and T. (1935b). F. (1949). Capitalism. London: Harper Collins. 1850–1890: a study in technological change and its diffusion’. G. or when the state sues the monopolist. N. (1970). B. Liverpool. C. A. Walmsley: United Alkali Company. P. L. Oxford: Oxford University Press. N. The American Ice Harvesters: A Historical Study in Technology. Harmondsworth: Penguin. NY: FO Willholft. (1976). M Phil Thesis. D. The Struggle for Supremacy. K. C. W. London: Longman. if the sailing ship effect is likely to be a rarity in the wider economy this implies a simplification of the strategic choices of firms facing a substitution threat. Essays on a Mature Economy: Britain after 1840. (1984) America’s Icemen – An Illustrative History of the United States Natural Ice Industry 1665–1925. Reindustrialisation and Technology. Administrative Science Quarterly. ‘The shift from sailing ships to steamships. (1935a). H. S. MA: Harvard Business School Press. London: Methuen. (1971). (1982). H. G. C. these ‘suppressed’ technologies may be more vigorously exploited only when an alternative. J. R. Boston. and M. CA: University of California Press. C. If Dunford’s many examples are valid. (Eds). The Electric Lamp Industry. 2nd edition. M. (1971). The Story of the Rotor. ‘Yardsticks for Victorian entrepreneurs’. Socialism and Democracy. ‘Timing technological transitions’. R. and Moore. Berkeley. (1949). London: Methuen. M. The Chemical Industry during the Nineteenth Century. (1953). O. Revolution in Miniature. ‘The suppression of technology as a strategy for controlling resource dependence’. 2. E. and Moore. ‘Strategic responses to technological threats’. L. 1987). J. N. (Ed. Readings in the Management of Innovation. threatening technology emerges that is uncontrolled by the monopoly firm. In Tushman. C. O. In McCloskey.  suppression’ (Dunford. W. W. and Z. (1992). S. Academy of Management Executive. (1982). Cambridge: Cambridge University Press.906 . © Blackwell Publishers Ltd 2002 . 6. R. C. (Eds). The Sociology of Invention. (Ed. D. (1996). L. A. D. and if so whether or not to ‘switch’ to the threatening technology. (1985). E. J. J. H. K. Mt Vernon. G.  A. (1958). Readings in the Management of Innovation. (1926). (1988). 2nd edition. ‘How established firms respond to threatening technologies’. (1987). Science Policy Research Unit. S. F. M. F. and S. London: Harper Collins. 32. In McCloskey. L. In Tushman. They must decide whether or not to exit the established technology.). R. Isambard Kingdom Brunel. Finally. R. C. A.