From 1875 the question was no longer one of the introduction of self-registering instruments to major observatories but their complete mechanization and the extension of registration to substations. Having accepted self-registration, meteorologists turned their attention to the simplification of instruments. In 1904 Charles Marvin, of what is now the U.S. Weather Bureau, brought the self-registering barometer into something of a full circle by producing an instrument (fig. 14) that was nothing more than Hooke's wheel barometer directly adapted to recording.[32] But this process of simplification had been accomplished at a stroke, about 1880, with the introduction by the Parisian instrument-maker Jules Richard of a self-registering barometer and a thermometer combining the simplest form of instrument with the simplest form of registration (see fig. 16). This innovation, which fixed the form of the conventional registering instrument until the advent of the radiosonde, seems to have stemmed from a source quite outside meteorology--the technology of the steam gauge. Richard's thermometric element was the curved metal tube of elliptical cross-section that Bourdon had developed several decades earlier as a steam gauge. Pressure within such a tube causes it to straighten, and thus to move a pointer attached to one end. Bourdon had opened it to the steam source. Richard filled it with alcohol, closed it, and found that the expansion of the alcohol on heating caused a similar straightening. His barometric element was a type of aneroid, which Hipp had already used but which Richard may have also adopted from a type of steam gauge. For a recording mechanism, Richard was able to use a simple direct lever connection, as the forces involved in his instruments, being concentrated, were not greatly hampered by friction.[33] By 1900 these simple and inexpensive instruments had relegated to the scrap pile, unfortunately literally, the elegant products of the mass attack of observatory directors in the 1860's on the problem of the self-registering thermometer and barometer.[34] Conclusions In view of the rarity of special studies on the history of meteorological instruments, it is impossible to claim that this brief review has neglected no important instruments, and conclusions as to the lineage of the late 19th century instruments can only be tentatively drawn. The conclusion is inescapable, however, that the majority of the instruments upon which the self-registering systems of the late 19th century were based had been proposed and, in most cases, actually constructed in the 17th century. It is also evident that in the 17th century at least one attempt was made at a system as comprehensive as any accomplished in the 19th century.