Book excerpt: This paper introduces a unique historical data set of more than 8,000 British and American innovations at world fairs between 1851 and 1915 to explore the relationship between patents and innovations. The data indicate that the majority of innovations - 89 percent of British exhibits in 1851 - are not patented. Comparisons across British and U.S. data also show that patenting decisions are unresponsive to differences in patent laws. Cross-section evidence suggests that high-quality and urban exhibits are more likely to be patented. The most significant differences, however, occur across industries; inventors are most likely to use patents in industries where innovations are easy to reverse-engineer and secrecy is ineffective relative to patents. In the late 19th-century, scientific breakthroughs, including the publication of the periodic table, lowered the effectiveness of secrecy in the chemical industry. Difference-in-differences regressions suggest that this change resulted in a significant shift towards patenting.