When Boulton and Watt first began to introduce their rotative steam engines into manufactories, about 1784, they had to commit to a certain power their machine would deliver. Boulton & Watt wanted to express this power in such terms as would be readily understood by the persons who were likely to want such engines. The machinery in the great breweries and distilleries in London was then moved by the strength of horses, and the proprietors of those establishments, who were the first to require Mr. Watt's engines, always inquired what number of horses an intended engine would be equal to.

In consequence, Mr. Watt made some experiments on the strong horses employed by the brewers in London, and found that a horse of that kind, walking at the rate of 2½ miles per hour, could draw 150 pounds avoirdupois, by means of a rope passing over a pulley. A least 5 different weights for a pound existed at the time in England. According to Watt's calculation 33 000 pounds were equal to 528 cubic foot of water, which makes 1 avoirdupois pound equal to 0.4531kg. Today, 1 avoirdupois in the metric system is assumed to be approximately 0.4536 kg, an error of aprr. 0.1%, which can be neglected. So Watt defined one horse power as the power to raise up the weight of 150 avoirdupois pound with vertical motion, at the rate of 220 feet per minute. As the steam engines in the beginning were mainly used to raise water from mines the horse power is often expressed as the power raising 33 000 pounds of water one foot high in a minute. [1].

According to a legend this high number is thanks to a brewer in London, a canny Scott, who chose his strongest horse to present it to Watt and was driving it eight hours to the limit of its endurance [2].  Fact is that Watt had chosen a rather high number for his horsepower as he did not want to be accused that his engines performed worse than announced. So he had some margin if the steam engine's performance would deteriorate with time.

Smeaton and other engineers made many observations on the work actually performed by horses when working regularly in mills, and the results seem to show that 22 000 lbs., raised at the rate of 1 foot per minute, may be taken for a real horse-power, or as the exertion that a good horse can overcome with so much ease as to continue to work for 8 hours per day. [3]. It was however common to have horses working in shifts of 2 or 3 hours, and let them rest for 2 shifts, which improved their performance above the 22 000 lbs raise by 1 foot per minute.

In Germany in a similar experiment, a horse raised 75kg in 1 second to a height of 1m. Probably the real figure was rounded to 75kg to have a round figure. This is the reason why one metric horse power defined as 75kgm/s is slightly smaller than 1 imperial horsepower of 76.04kgm/s. In SI units the metric horsepower is 735.5 Watt and the imperial horsepower is  745.7 Watt. In this case study I will use the imperial horse power as the historical documents mostly use the imperial horse power and thus the historical results are easy to compare. Still most of us are used to horse powers from the car engines, although with electrical cars this may change. 


[1] John Farey, page 439

[2] Popular Mechanics, 1912, page 394