I disagree with Al Moore's method of breaking the specimen.
To me, the fillet break test is a great opportunity to really let the welder feel the difference between a good weld and a bad weld. I define a bad weld as one where the penetration does not reach the root. I define a good weld as one where penetration does reach or go beyond the root.
If you make the welder break the coupon with a steady force applied through a lever, like a beam turner (Beam Peavee, Turning bar, Etc.), he will really feel the difference that 1/32 inch of bridged root makes when he finally gets the root right! I have given hundreds of these tests, (usually 5 or 9 to each welder), and very few get the penetration to the root on the first try. The effort required to break the coupon the first time,when the root is bridged by that 1/64th or 1/32", is half what it takes when the welder finally gets penetration to the root. Even the dumbest of welders feel the difference.