I've read how to do this by doing test runs, using the width and distance spread vs weight spent. I don't really like the idea of wasting a bunch of fertilizer doing this, and 100 times so on expense seed. 
You can fairly quickly, and with little waste, find the effective spreading width with your material. Distance spread is simply a factor of time. So using that why couldn't one just pull the spinning disc off and just let fertilizer drop into a bucket to be weighed.
Example;
50' spreading width
10mph (14.67 feet per second)
=733.5 sqft/s = 1 acre coverage every 59.38 seconds, round up to 1 minute.
Now if you know you need to spread 500# fertilizer per acre, can you not just pull the disc and see which setting drops 500# in a minute (or 250 in 30 seconds, or 50# in 6 seconds). There should be no difference in amount through the gates while actually spreading with the disc or dropping straight down. It's not like the spinning disc causes a backup or slows the rate or anything.
Practical tests are always going to trump, but this should output very close to a practical test, maybe even more accurate since it removes a factor of operator error.
Am I crazy?
You can fairly quickly, and with little waste, find the effective spreading width with your material. Distance spread is simply a factor of time. So using that why couldn't one just pull the spinning disc off and just let fertilizer drop into a bucket to be weighed.
Example;
50' spreading width
10mph (14.67 feet per second)
=733.5 sqft/s = 1 acre coverage every 59.38 seconds, round up to 1 minute.
Now if you know you need to spread 500# fertilizer per acre, can you not just pull the disc and see which setting drops 500# in a minute (or 250 in 30 seconds, or 50# in 6 seconds). There should be no difference in amount through the gates while actually spreading with the disc or dropping straight down. It's not like the spinning disc causes a backup or slows the rate or anything.
Practical tests are always going to trump, but this should output very close to a practical test, maybe even more accurate since it removes a factor of operator error.
Am I crazy?