real RS232 signals use +8 Volt and -8 Volt as signaling levels
It"s a pedantic point, but real RS232C uses +/- 12V. However the receive threshold is at +3V, so it will inter-work with a lot of different voltages. For example the serial port on the BBC Micro was RS423, which used +/- 5V. Most of the time it worked OK as if it was RS232.
However it will not inter-work with the RaspPi, which doesn"t have the right drivers or signal sense, and wont cope with a 12V incoming signal.
Yes, I didn"t want to make this more complex than it already was. Most RS232 driver IC"s run of 5V and use an internal voltage doubler/inverter so the actual voltages are often + and – 10Volt.
Another important point is that RS232 drivers are inverters, so even if you use resistor divider to convert +12V to +3 volt it would not work, you will need at least a single transistor, switched as logic inverter. with a base resistor (and probably a diode between base and collector to protect the transistor against -12V on its base) and a pullup to 3K3.
A similar transistor inverter can be used on the TXD line.
I object to the remark that the R-PI only has an RS232 interface in an academic sense!
its much easier/cheaper to add a RS232 port using the available UART than any other way (USB to RS232 device). Also, the CTS and RTS handshake signals are also available using other GPIO pins, Board designs have already been published, and use only a few cm2 of board space, with the DB-9 connector taking the most space. Details are published elsewhere on this forum. Its definitely doable.
Where the UART shines, and will probably be used the most for, is to communicate directly (without RS232 level conversion inbetween) with another microcontroller, such as the gerboard does to connect an ATmega (as used in arduino"s) to the R-PI.
Also, the original comment on the farnell site was that the R-PI has a "serial port", which is strictly seen true, it just that a serial port isn't always using the RS232 protocol.