OK Web, just for you.
Take depth as a series of layers, 100m thick.
The top layer, layer (1) has a Temp of T and the bottom layer is in steady state with the bottom layer, layer (21), that is at T-15.
Now the overall heat flux is always from layer (1) to layer 21.
Now perturb the system by the addition of heat to layer one, raising its temperature. In pre-steady state there will initially be a rise in temperature, and heat content, in layer (1), then heat will begin to probe the lower levels; the profile, for a constant rise in surface temp, will have an exponential.
At steady state the layers near the surface will have a larger change in heat content than those at the bottom.
If we set layer 21, >2000m, as a true heat sink, then its temperature will be unchanged. If we define ‘unchanged’ as less than 1% of the heat change in the surface layer we get an exponent of 0.2 layer-1.
Therefore heat contents of layers are
0-300 =1
0-700 =1.6
0-2000 =2
and so 300-700m =0.6
Therefore the ratio of 0-300:0-700 is 1:1.6.
The 0-300m heat content should be almost exactly half of the 0-2000m heat content.
However, this is the steady state approximation, in pre steady states the ratios will always be lower, so
0-300:0-700 is 1:<1.6
0-300:0-2000 is 1:<2
If we take the figure that Judy has placed at the top of the tread, from 2000 until the end of the graph, trying to get the biggest 0-300 and smallest 0-2000, we get;
0-300m = 5.2 * (10^22J)
0-700m = 9.6 * (10^22J)
0-2000m = 13.2 * (10^22J)
0-300:0-700 is 1:1.84 compared with 1:<1.6
0-300:0-2000 is 1:2.53 compared with 1:<2
Your lines do not match the data for the 0-300 and 0-700, you had to cheat by making your 0-300m overshoot and the 0-700m undershoot.
http://img819.imageshack.us/img819/7860/ohc.gif