I could be wrong, but I think the logarithmic response is in the wrong direction. The response of the LM3915 is most sensitive at low voltages where the RC charging voltage would be moving most rapidly. Thus the first few LEDS would light very rapidly with a signficant slow down at the end.
If you charged the cap to the high voltage level and monitored the RC discharge, then I think the lights would turn off in a linear fashion. You could get a sequential linear turn-on in this fashion if you added inverters to each LED output.
Or is my thinking backwards?