Zhang et al. reply

replying to J. Luo et al. Nature 501, 10.1038/nature12589 (2013). We agree with Luo et al.1 that the magnetoresistance effects that we reported2 were dependent on the method used to measure them. The reason that there is a difference in the results depending on whether method 1 or method 2 is used (adopting the measurement notation of ref. 1) is that there are two voltage-stabilizing diodes in the Keithley 2400 instrument we used. We were unaware that when this instrument was used both as current source and voltmeter, one diode connected the input port of the current source to the input port of the voltmeter, whereas the other diode connected the output port of the current source to the output port of the voltmeter. The diodes caused a crossover of the Hall coefficient from negative to positive when the instrument was used to conduct a Hall measurement in this configuration, leading us to propose an invalid mechanism for the abnormal magnetoresistance. Therefore the mechanism we proposed2—minority injection and an induced p-n boundary—does not provide a correct explanation for the observed geometry-enhanced magnetoresistance. Although such a mechanism does not operate in our samples, we note that a p–n boundary could still enhance magnetoresistance in certain circumstances according to our and others’ theoretical calculations and experiments2,3,4.