To add a little bit more to the confusion:
I just wired up a SR227 Split Receptacle Module (White, Date code sticker 00A03) in a handy box I use for testing things like this.
I plugged in a 100 Watt incandescent lamp and verified correct On/Off operation by sending X10 powerline commands.
I turned the lamp Off and On repeatedly using the rotary switch in the base of this lamp, but the module did NOT switch on.
I then unplugged the lamp, and with the lamp switch in the On position plugged it in again. This time the module DID switch On. I repeated this test several times with the same result MOST of the times. I appeared that jiggling the plug a little when inserting rather than ramming it straight in was better at getting the module to switch on.
I then tried the same test with a 1500 Watt GE Toaster Oven. The module would NOT switch On if I turned the toaster over On with its own switch, but it would switch On when I plugged in an already-turned-on toaster oven.
Finally, I plugged a plug strip into the module and retested plugging in the lamp, but this time into the plug strip. The results were the same as plugging the lamp directly into the module, including the effect of jiggling the plug. This particular plug strip has a neon bulb pilot light, and so long as there is no load plugged in, the neon bulb remained On when the module was Off, indicating the module is attempting to sense the load.
It would appear that the little bit of arcing when a plug is jiggled into the socket is more effective than a quick clean insertion in causing the module to switch On.