Constraints on temperatures in the deep interior can be obtained from measurements of the pressure (P) - temperature (T) slopes of the phase transitions responsible for the major seismic discontinuities. For example, the inner core-outer core boundary (IOB) at 5150 km depth (329 gigapascals (GPa)) separates the liquid outer core from the solid inner core. Measurements of the melting curve of iron, when extrapolated to IOB pressures, establish an upper bound to the Earth's temperature at this depth. A number of studies have reported measurements of the melting curves of iron and iron alloys in the laser-heated diamond anvil cell [ Williams et al., 1991; Knittle and Jeanloz, 1991; Boehler 1992, 1993; Saxena et al., 1994].
For pure iron, the melting curve measured by Williams et al. [1991]
to 105 GPa is in strong disagreement with those of Boehler [1992, 1993]
to 200 GPa and Saxena et al. [1994] to 150 GPa, with the melting
temperature of the former being 
1100 K higher at 100 GPa. By
extrapolating his measurements to core pressures, Boehler [1993]
estimated the temperature of the Earth's IOB is 4850 
200 K and the
temperature at the core-mantle boundary (CMB) at 2891 km depth (135 GPa)
is about 4000 K. Recent shock compression experiments to 340 GPa [ Yoo
et al., 1993], however, yield a melting temperature of iron at the IOB
about 2000 K higher (6830 
500 K), and in general agreement with
earlier shock melting experiments as well as static
measurements of Williams et al. [1991].
An additional problem that has not yet been explored in detail is the
effect on melting temperature of the wide variety of possible alloying
components in the core (e.g., H, O, C, Si, S, and Ni).
There has been sharp disagreement over the melting behavior of FeO.
Knittle and Jeanloz [1991] measured the melting curve of FeO
to 102 GPa, and inferred that the melting temperature of FeO exceeds
that of Fe by 1000-2000 K at the CMB pressure.
Assuming oxygen is a major alloying component in the outer core, a CMB
temperature of 4800 
500 K was inferred from this data.
Boehler [1992, 1993] and Shen et al. [1993] obtained
significantly lower melting temperatures (by 
800 K at 50 GPa)
for FeO than Knittle and Jeanloz [1991], and Boehler [1993]
found no measurable difference in the melting temperature of Fe and Fe-FeO
alloys above 60 GPa. Boehler's [1992] measurements of the melting
curves of FeS and 
to 50 GPa show strong melting point
depression relative to Fe. Shock temperature measurements on an
Fe-19%Cr-9%Ni alloy reveal that this material melts at 5800 
300 K at 250 GPa, providing constraints on the effect of transition
metal alloys on iron melting temperatures [ Gallagher et al.,
1994].
Using dislocation theory, Poirier and Shankland [1993] calculated
a melting temperature of 6160 
250 K for pure iron at 330 GPa, and
estimated that the temperature of the IOB is 5160-5660 K. In a review of
experimental data that includes a possible new high-temperature phase of
iron above 200 GPa, Anderson [1993] concluded that an upper bound
to the melting temperature of iron at the IOB is 6500 K, and taking account
of plausible depression due to light elements, an upper bound for the
central temperature of the Earth is 5700 K. Further discussion of the
iron phase diagram can be found in papers collected in Schmidt et
al. [1994].