Puchkov and Timusk Reply: Combining our own optical measurements and those of several other labs we showed that infrared conductivity spectral weight (SW) increases with T c in underdoped high-T c cuprates but saturates above the optimal doping. In their Comment Kendziora et al. argue that a thin-crystal transmission measurement, in an unusually highly doped sample of Bi2212, suggests an increase in SW in contradiction with our finding. The results of Kendziora et al. [1] are interesting and should be pursued further, but their interpretation is not straightforward for the following reasons.First, in Fig. 1 we plot the high-frequency ( Fig. 1 in the Comment [1]) and the low-frequency transmission results (Ref. [2]). There is a large disagreement between the two data sets. Both sets are marked by the authors as obtained on T c 60 K (oxygen-annealed) and T c 85 K (air-annealed) crystals, implying that the material (if not the actual crystals) was the same. No single multiplication coefficient, such as crystal thickness, can bring the two sets of data in agreement with each other. If one multiplies the high-frequency curves by different coefficients in an attempt to remove the discontinuity, one finds that the transmission doping dependence can be accounted for by a constant plasma frequency and a decreasing with overdoping scattering rate-in agreement with our findings. Next [3], following the authors of the Comment, we obtain a ratio of the dc resistivities at T 300 K from the transmission data: ͓t͑0͒ air ͞t oxygen ͑0͔͒ 1͞2 r air dc ͞r oxygen dc 1.22. Contrary to what the authors claim, this ratio is in disagreement with the directly measured ratio 1.73 [2]. The Hall effect results (Ref. [5] in the Comment) are irrelevant since no overdoped samples were measured. A fit [2,3], similar to the one used by the authors of the Comment, gives a 50% increase in the scattering rate of the oxygen annealed sample. This is contrary to a general trend that overdoping decreases a scattering rate [4,5], and suggests a large increase in an impurity concentration.We do not wish to discard the results obtained on La 22x Sr x CuO 4 as irrelevant since the relation between T c and x is very well established for this system. For example, a measured decrease of T c with overdoping from 27 to 22 K (more than 20%) to 0 K produced less than a 2% change in SW. No significant increase in the (low frequency) SW can be seen in the results by Prenninger et al.Other experiments support our observation that the SW stops increasing at optimal doping. Fukuzumi et al. suggested that while the carrier concentration increases in the underdoped regime, it decreases in the overdoped regime [7]. Photoemission results suggest that the Fermi surface volume increases with doping in the underdoped regime but decreases in the overdoped regime [8]. In fact, since the (holelike) Fermi surface of the optimally doped FIG. 1. The high-frequency and the low-frequency sets of transmission curves by Kendziora et al. Also shown are our fit to the low-frequency result...
