Eli Lilly’s mass production of insulin required many rabbits. One estimate is that Eli Lilly performed over 100,000 rabbit assays during the last six months of 1922.[McCormack, Discovery and Manufacture of Insulin]
The standardization of the unit of insulin presented several problems. It relied upon a non-standard itself, a rabbit. Even if you had standard rabbits, the standard relied upon their diets. If you standardized the rabbit’s diet, variability still existed.
The Toronto insulin committee tried to establish a standard unit, despite these complications. They first established that the rabbit should: weigh 1 kilogram, have no food for 24 hours before testing, and have hypoglycemic convulsions within three hours of the insulin injection. When those administering the insulin noted that different rabbits had consistently different responses to the insulin, the Insulin Committee decreed that blood sugar levels should be measured at three different time points: 1 ½ hours, 3 hours, and 5 hours, after injection of insulin. Specifically, they decided that the number of units per cubic centimeter would be determined by the formula, a/b x w/c x ½ where a was the difference between normal blood sugar and the average blood sugar at the three time points, b was the difference between normal blood sugar and 0.045 percent, w was the weight of the rabbit in kilograms, and c was the number of cubic centimeters injected.[sinding, p 247]
Despite the detailed specification of a physiological standard from rabbits, some found it still too variable. Clowes, the director at Eli Lilly, found the physiological standard almost useless. In a letter to Macloeod dated January 8, 1923 [sinding, p 257], he suggested just basing initial doses on the amount of pancreas from which the insulin batch was purified. Then perform initial tests in the clinic to refine the calibration. He lamented that they were using 100 rabbits per day on these tests that were not adding any value.
Doctors who treated diabetic patients had always rigorously tracked their patient’s inputs and outputs. They knew exactly how many carbohydrates their patient ate, and how much each gram of carbohydrate raised their patient’s blood sugar levels. Similarly, they routinely measured how much insulin each patient received and how much the blood sugar dropped as a result. Tracking these values were a key part of developing a treatment plan for a diabetic patient.
Clowes was in constant contact with insulin users. They sent him letters describing the quality of each batch of insulin. Clowes found these letters more useful for determining the activity of each batch of insulin than the rabbit assays.