Why is char a numeric type in Java, and what breaks when you treat a String as a sequence of chars for non-Latin text?

char is an unsigned 16-bit UTF-16 code unit, so it participates in arithmetic and 'A' + 1 yields 66.

The trap is that characters outside the Basic Multilingual Plane (emoji, many CJK extensions) are encoded as surrogate pairs spanning two chars, so str.length() and charAt() count code units, not code points. For correct counting or iteration use codePointCount, codePoints(), or String methods that are code-point aware.