Brass vs Steel?
My first thought is I really do not see any difference in the difficulty level for using either of those two materials. The very same drill bits that can drill holes in brass will also drill them in steel. In fact, due to the tendency of standard twist bits to grab in brass, drill in steel is perhaps easier and prone to fewer mistakes. I have a complete set of drill bits that are resharpened to not grab in brass or plastics.
I need to make some index plates to cut watch or clock gears on my bench lathe.
That seems to indicate that you are going to use direct indexing. That is, the plates will be used without any kind of indexing or dividing head. And it is for gears for clocks and watches so a decent amount of precision will be needed. I am not going to go beyond that statement as I do not know the consequences if a tooth on a watch or clock gear is off by any particular amount. At the very least, you would not want the movement to stop due to bad gear teeth.
Brass in general is softer and will wear more rapidly than steel. That may not be true for every combination of alloys, but in general it is probably true. So we must consider just how much use these index plates will get and just how much wear is a tolerable amount before the plate is no longer useful and must be replaced. I can't answer that, but you should. And even with a completely accurate answer, it will still be hard to say just how long each metal would last.
But I would like to know just how you propose to make these plates. If you have a rotary table or an indexing head, then you have the perfect way to make them. I have often written about just how some almost perfect hole circles can be made with those devices. This is due to the fact that the worm mechanism in those devices functions as an "accuracy amplifier". With a 40::1 worm a second generation plate will have errors that are only 1/40 as large as the first generation plate used to make it. And a third generation plate will be another 40 times better than the second generation one. So, by making three generations of plates, the first one can be laid out in the most crude manner possible and by the time you make the third generation plate from it, the errors will be below the level of the errors in the rotary table or indexing head that you use.
If you are going to use the plates for direct indexing, then a third generation one would be needed. But if you are going to do the actual work with a rotary table or indexing head, using the plates you make, then the second generation plate is all you need because the second amplification of the accuracy will take place when the plate is used and your parts will be as accurate as your rotary table or indexing head is capable of.
However, I have seen many other methods for making indexing plates. Some of these use rather questionable pattern pieces, like circular saw blades. Others are just laid out by hand, using dividers. Many people do it this way, but in my mind, it is questionable. It is especially questionable for direct indexing. And it is especially questionable if you are going to be making gears.
Still other plates are made on milling machines with DROs (digital read outs) using coordinates generated by those DROs or by a CAD program. These can be satisfactory if care is used in drilling the holes: and that is another subject. Still more have been made with milling machines with CAM or computer assisted machining capability. Those are also very acceptable, again with care in how the holes are drilled.
But back to the basic question. I would have to say that steel would be my choice and for indexing plates I would probably choose a harder alloy, like 4140, prehardened. That alloy will give longer service while not being excessively difficult to work with. If you decide to go with brass, I would recommend a hard brass, like Marine Grade 485.