Because I know a lot more about RPDs on Indian Cents than Shield Nickels, I've been doing some comparing between the two. Today, I just happened to see two RPDs on an IHC and a Shield Nickel that look so alike, they could almost be "brothers". Granted, the digit shapes and spacing differ, but the actual RPDs are strikingly similar. This suggests to me that both IHC and Shield Nickel dies were dated using the same process. In both cases, there are similar details worth noting. When a date is repunched in a die, there's often a rotational aspect. In other words, the digits have different degrees of spread, due to this rotation. And typically there is some separation between the underlying punch and the final digit. You see that below as a gap between the top of the 67 and the underlying date. This is caused when mint workers attempt to polish away the mistake, but leave traces of underlying digits--such as where the top stroke of the 6 is reduced to a faint line. I have seen exceptions too, such as where the first digits are pressed so deeply as to change the shape of final digit--like as the "9" in the second picture. In this case, the outer size of the digit is expanded, but the inner loop will be made smaller. Of course, bronze and CuNi coins strike up differently, with different degrees of machine doubling seen. This should also be taken into account when evaluating a coin for an RPD. I admit, I need to look at a lot more Shield Nickels to correctly distinguish RPDs from more common MD. Top: 1867/67 Snow-1 IHC. Bottom: Shield Nickel in PCGS slabbed as "minor variety"--although the RPD is not minor at all. Some repunching actually changes the shape of the final digit on the 1889 Snow-4. The colored overlays show the spread between repunchings.
I see the nickel bug has nipped at you. It could be possible that the same punches were used on both I say this not first hand but the script are about the same. Plus the time period at the Philadelphia mint anything could of happen. Remember we are dealing with humans and human error. Plus the Philly mint where it was 5 o'clock all the time.
