Understood. Being conservative is always the safest path. I'd say that, in the danger zone, velocity and pressure MAY stop acting together. Increased pressure necessarily leads to increased velocity UNLESS that pressure does not actually act upon the base of the bullet (as in the case where it instead leaks past the brass, blows by the bullet to bore seal, escapes via the cylinder gap, or breaches the primer). If additional powder fails to produce additional velocity, we might conclude that the invisible intermediary, pressure, became higher than the system could contain. That would certainly be the time to stop and drop a half grain. However, it is also plausible that additional powder simply did not create additional pressure because there was too much present to burn before the bullet escaped and uncorked the little bomb we call a pistol. Again, erring in favor of keeping all your digits is wonderful advice. For the benefit of any less experienced reloaders, I would simply add that a situation such as you observed is a good time to consult multiple sources of data. Presumably you started with a published load from one manual. Consult some others. While we do not have direct pressure data available at our bench, the powder companies have extensive test info and we can compare our results to their recommendations for a clue as to whether we are indeed playing with fire. Likewise, examine the burn rate chart. Is the powder you are using similar in burn rate to those powders used in published loads? If your powder is slower than what is commonly used, expect to see something like Wobbly noted in the .38 loads, where velocity plateaus and maybe muzzle flash gets ferocious. If you are using faster powders, expect that you may not be able to reach the published velocity before seeing pressure signs. Fast powder is lovely for creating mouse fart loads but usually reaches the pressure limit fast and then fails to sustain it long enough to accelerate the bullet to higher speeds. Sticking within published data is the safest course, but even then we can encounter results that are confusing without the benefit of a ballistics lab to provide an xray into what's happening. When you get those weird results you can hypothesize all you want about the cause, but don't ignore the availability of additional published resources to help you determine whether you are being safe.
In this case, Hodgdon says max load for 148 grain lead is 4.0 grains. They got about 950 fps, but in a 7.7" barrel. Your results seem about as expected with the shorter 6" barrel. Its also likely that the cylinder gap accounts for some loss if the data was run in a test barrel.
38 special is a fun example because we have +p and 357 load data to examine as well. In the +p listings, it shows 4.9gr hp38 producing just 25 fps more (976fps) and pressure "only" about 2k higher than in the max load for 38 special. I take this as confirmation that 231/hp38 just cannot burn fast enough to make more pressure in a short barrel. The .357 data is run in a 10" barrel and we see that jacking the charge to about 6 grains does produce notably more pressure, given the longer barrel in which to burn.
Sorry if this went off the rails into complexity but I know there will be folks who can benefit from a reminder to use the info the powder companies spent so much money compiling. When your results start to look funky, stop loading and start studying.
Happy New Year all!
Sent from my SM-G960U using Tapatalk