It's a CA3081 NPN transistor array.
It worked for me! The mystery chip is definitely the CA3081's twin brother.
To prove that they are equivalent, after an initial test with the ohm-meter, I tried the following simple tests:
Conditions: substrate (pin 5): to GND. All emitters (pin 15): to GND.
Test 1:
LED with 180 ohm resistor in series from each collector to +5V (outputs active one at a time):
the LED lights up when the corresponding base is biased using 2.7 Kohm to +5V.
Measured Vce sat: about 150 mV.
Test 2:
LED with 560 ohm resistor in series from each collector to +12V (outputs active one at a time):
the LED lights up when the corresponding base is biased using 10Kohm to +12V.
Measured Vce sat: about 150 mV.
Test 3:
300 ohm, 12V relay from collector at pin 14 to +12V, with suitable reverse-biased protection diodes to both +12V and GND:
the relay gets energized when the base at pin 13 is biased using 10Kohm to +12V.
Measured Vce sat on pin 14: about 350 mV.
Therefore, with the appropriate external components, a 733W00186 could replace a ULN2000 in non-critical (i.e. lower output current) applications.
The bottom line is that I have turned a bagful of junk lying in my drawer into some useful components, and you deserve most of the credit.
Many thanks to all the forum members who have been taking part in this thread.