Do estimates of numerosity really adhere to Weber’s law? A reexamination of two case studies