A description of a n improved radia l heat flow technique for measuring the thermal conductivity of so l ids in the range -57 to 1 1 0 0 ~ ~ is presented. The technique yielded results with a probable accuracy of f1.5% and a reproducibility of f0.1% in th is range. Meaningful measurements were limited to 1 1 0 0 ~ ~ by P t vs Pt-10% Rh thermocouple instabil i ty, although the apparatus was structurally sound to 1 4 0 0 ~ ~ . The thermal conductivity of polycrystall ine UO was measured from -57 to 1 1 0 0 ~ ~ and reveals a maxi2 miim in k near room temperature caused by the decrease of the la t t ice speci f ic heat. The thermal res is tance , l / k , shows a l inear dependence with temperature from 200 to I.OOOOC, which is expected for a n insulator well above i t s Debye temperature. The s lope of the l/k-temperature plot is 0.0223 cm/w, and th is is independent of impurity content, although the intercept is sens i t ive to impurity content. The thermal conductivity of polycrystall ine ' Armco iron was measured between 100 and 1 0 0 0 ~ ~ and was found t o be within f 2 to 3% of the bes t values reported in the literature. The temperature dependence of the thermal conductivity of iron is largely controlled by the e lec t ronic contribution, which was deduced from elec t r ica l conductivity measurements. The thermal conductivity of iron can be' represented by four linear equations for the temperature ranges 0 to 436, 136 to 786, 786 to 910, and 910 to 1 0 0 0 ~ ~ . A s lope change of, 30% a t 436% may be coupled to a ii~it~imurn in the thermoelectric power of iron near th is temperature. A minimum in the thermal conductivity occurs near 786OC and is associa ted with the Curie transformation. A 4% decrease in the thermal conductivity was observed a t the a-y transformation (910°C), and th is is associa ted with a change in the la t t ice contribution to the to ta l thermal conductivity.