(eng): We introduce a new version of a piecewise linear histogram recently introduced and studied in the literature. Our version is a bona fide density estimator. It is proved to be strongly consistent and tending to zero with the same expected L1-error rate as the classical histogram. However, as simulations demonstrate, for simple sizes between 20 and 10000, it achieves smaller errors than the classical histogram and the formerly introduced piecewise linear histogram.