Comments on oxygen isotope geothermometry