cluster cutoff bug?

I use mothur 1.33, and when i use cluster command, the cutoff changed to a small number(even with parameter hard=true).
when i set cutoff to 0.03, i only get the unique label, when i set cutoff to 0.10, i can get labels: unique/0.01/0.02/0.03.

mothur > cluster(column=dataset_4444.dat,name=dataset_4433.dat,method=average,cutoff=0.1,hard=true,precision=100)
Reading matrix: |||||||||||||||||||||||||||||||||||||||||||||||||||||

changed cutoff to 0.0332595

label numOtus
unique 79527
0.01 39448
0.02 15304
0.03 10731

In the wiki, it says “By default the cutoff parameter is set to cutoff + (5 / (precision * 10.0)).” column parameter is my dist file, and there are distances bigger than 0.15. I want to know how the cutoff changed itself?