Electronic – Why does an oscilloscope show a different voltage from a signal generator

I have a Rigol DS1052E oscilloscope and a Rigol DG1022 function generator. I have just changed the oscilloscope fan for a silent one, as I found, as is commonly reported, that the original fan noise was very irritating. That seems successful and is the only modification I have made to the oscilloscope.

Testing the oscilloscope as a check, I connected the signal generator to it with a plain BNC cable (both ends BNC), and set the signal generator to a 5 volt peak-to+peak 1 kHz sine wave, and the voltage was confirmed by a digital voltmeter showing 3.67 or so V RMS. However, the oscilloscope shows 10 V p/p, and this also happens for a 100 kHz sine wave. Testing other amplitudes, it is not always precisely two to one, but it is round about that ratio.

Why does the oscilloscope show twice the voltage as the signal generator? I hope it is something trivial that I have overlooked, and desperately not an "undocumented hardware modification", as we used to say in the lab when we dropped something and it no longer worked.

Well I need not have panicked. @Sephro has the solution. Thank you to him especially, and to the others for the helpful comments. I would have also accepted @alex forencich's version if I was allowed to. I was beginning to approach that answer myself, since I discovered that inserting a 50 ohm parallel load on the output of the sig. gen. cured the problem, I have done as @Seph suggested and set the sig. gen. output to "High Z" and that, too, makes the voltages shown coincide.

So, my extended explanation, from the answers given, is:

I see now that, to generate 5 V across a 50 ohm load, with its internal resistance also 50 ohms, the sig. gen. actually generates 10 V of which, because the two 50 ohms in series act as a voltage divider, only 5 V appears at the output terminals, and that is the value that is displayed on the sig. gen. panel. If a high impedance load such as an oscilloscope is attached, then the voltage divider is distorted, with a very small drop across the sig. gen.'s internal 50 ohms, and very nearly all across the 'scope input. The sig. gen. does not "know" this and still gives the same display as for the 50 ohm load. As @Seph says, there is a "High Z" setting for the DG1022 sig. gen. that (in effect?) replaces the internal impedance with a high impedance to match the load, so the voltage divider effect is restored.