The first point that is immediately clear is the complete

The first point that is immediately clear is the complete change in

the production structure of the mule mills between 1884 and 1890.

More specifically, one of the most representative examples of this

change is the increase in the output per spindle (nearly the same as

the increase achieved by all mills) over the period. As is also shown

by Discriminant Function A, variables such as power source, the

female worker ratio and yarn count are the variables that characterize

the differences in the production structures in these two years (unlike the results from univariate analysis).

Second, although the productivity of the mule mills improved greatly

towards 1890, the production structure of these mills was still inferior

compared with those mills that had introduced new ring frames a

little earlier. Discriminant Function C confirms that there is a large

distance between those two production structures. This distance is

mainly caused by differences in average count, per spindle productivity and operating days. These results can be interpreted as reflecting

the indirect influence of the technological superiority of the ring frame,

since the ring frame was capable of better operating results in the

same production system.

Third, the basic production structure of the various types of mill

had been more or less determined by 1893, although up to that time

there had been a gradual increase in per spindle productivity and a

reduction in the wage differential between male and female workers

across the industry as a whole. In short, between 1890 and 1893,

there were virtually no changes in the production structures of each

type of mill (Discriminant Functions B, E and H are not statistically

significant), excluding the fact that the proportion of ring frames in

the mills that used both types of machine rose from 55.3 per cent to

63.4 per cent.

Let us turn to the implications of these three points for the hypothesis that ring frames were adopted so that male workers could

be replaced by female workers. First, it is even clear upon examination 

of the figures for 1890 - the year immediately after ring frames were

adopted in large numbers - that the female worker ratio was certainly not the key factor which characterized the difference in the

production structure of mule mills and ring frame mills. A comparison

of the female worker ratio in ring frame mills (such as Kanegafuchi,

Hirano Boseki and Settsu Boseki), and the female worker ratio in a

mule mill (Naniwa Boseki) and dual mills (such as Osaka Boseki, Mie

Boseki and Temman Boseki) shows unsystematic differences in the

ratio between these mills.30 Even in the case of Osaka Boseki, it seems

natural to conclude that the female worker ratio was gradually rising

irrespective of the type of machine being used. In other words, the

reliance of the Japanese cotton-spinning industry on unskilled women

workers gradually became established after the switch to ring frames

and increased up to 1897. Thus, technological factors were not the

only reason for the reliance on women. Second, amongst the ring

frame mills of the 1890-3 period, the mills with a higher female worker

ratio (for example, Kanegafuchi and SenshfIBoseki) had smaller malefemale wage differentials and a greater intensity of operation (i.e.,

number of operating days and number of operating hours per day).

This tendency can be confirmed by the correlation coefficient matrix

and it indicates that the adaptation of each mill to market conditions

occurred after the adoption of the new spinning machine. It is now

necessary to return to examining the factors that allowed the mills to

make such a rational choice of technology: the choice of the ring

frame.

TOWARDS A CONCLUSION: ENGINEERS TAKING THE

LEAD IN TECHNOLOGY CHOICE

As can be conjectured from the above, the crucial reason for choosing the ring frame was that it was a new and powerful machine that

embodied the new technological innovations of the time and was

more efficient than the mule, particularly in Japanese market conditions. It is necessary, however, to make explicit some of the preconditions that made this choice possible and that so far have tended

to be overlooked.

First is the prerequisite that accurate information regarding the

relative efficiencies of the mule and ring frame was already known in

the Japanese spinning industry before 1889. This condition was shown