The violin bridge filter role was investigated using modal and acoustic measurements on 12 quality-rated violins combined with systematic bridge rocking frequency f(rock) and wing mass decrements deltam on four bridges for two other violins. No isolated bridge resonances were observed; bridge motions were complex (including a "squat" mode near 0.8 kHz) except for low frequency rigid body pivot motions, all more or less resembling rocking motion at higher frequencies. A conspicuous broad peak near 2.3 kHz in bridge driving point mobility (labeled BH) was seen for good and bad violins. Similar structure was seen in averaged bridge, bridge feet, corpus mobilities and averaged radiativity. No correlation between violin quality and BH driving point, averaged corpus mobility magnitude, or radiativity was found. Increasing averaged-over-f(rock) deltam(g) from 0 to 0.12 generally increased radiativity across the spectrum. Decreasing averaged-over-deltam f(rock) from 3.6 to 2.6 kHz produced consistent decreases in radiativity between 3 and 4.2 kHz, but only few-percent decreases in BH frequency. The lowest f(rock) values were accompanied by significantly reduced radiation from the Helmholtz A0 mode near 280 Hz; this, combined with reduced high frequency output, created overall radiativity profiles quite similar to "bad" violins among the quality-rated violins.