diff --git a/src/engine/__tests__/lexer.spec.ts b/src/engine/__tests__/lexer.spec.ts
index a70d420..8ad9da2 100644
--- a/src/engine/__tests__/lexer.spec.ts
+++ b/src/engine/__tests__/lexer.spec.ts
@@ -62,6 +62,15 @@ describe('lexer', () => {
     expect(values('$A$1 A$1 $A1')).toEqual(['$A$1', 'A$1', '$A1', '']);
   });
 
+  it('reads identifiers with non-ASCII letters (umlauts etc.) as a single IDENT', () => {
+    // a table/sheet name routinely contains "ä"/"ö"/"ü"/"ß" in this
+    // German-first app -- these must lex as ordinary IDENT characters, not
+    // throw, so an unquoted qualifier like "Übersicht::A1" actually works.
+    expect(types('Übersicht::B4')).toEqual(['IDENT', 'DBLCOLON', 'IDENT', 'EOF']);
+    expect(values('Übersicht')).toEqual(['Übersicht', '']);
+    expect(values('Größe_Prämie')).toEqual(['Größe_Prämie', '']);
+  });
+
   it('skips whitespace', () => {
     expect(types('  A1\t+\n B2 ')).toEqual(['IDENT', 'PLUS', 'IDENT', 'EOF']);
   });
diff --git a/src/engine/__tests__/workbook.spec.ts b/src/engine/__tests__/workbook.spec.ts
index 9ad1740..44203c1 100644
--- a/src/engine/__tests__/workbook.spec.ts
+++ b/src/engine/__tests__/workbook.spec.ts
@@ -90,6 +90,17 @@ describe('Workbook cross-sheet', () => {
     expect(val(wb, 's-kosten', 'A9')).toBe(800);
     expect(val(wb, 's-kosten', 'B9')).toBe(800);
   });
+
+  it('a table name with non-ASCII letters works as a bare, unquoted qualifier', () => {
+    // regression: "Übersicht" is the fixture's own table name, but nothing
+    // above ever refers *into* it by name (it only ever refers out) -- so
+    // this table has silently been unreferenceable from elsewhere the whole
+    // time (the lexer rejected "Ü" outright; see engine/lexer.ts).
+    const wb = fixture();
+    wb.setCell('s-umsatz', 6, 8, '=Übersicht!B2'); // G9, no quoting needed (no spaces)
+    expect(val(wb, 's-umsatz', 'G9')).toBe(5970);
+    expect(isError(val(wb, 's-umsatz', 'G9'))).toBe(false);
+  });
 });
 
 describe('Workbook error values', () => {
diff --git a/src/engine/lexer.ts b/src/engine/lexer.ts
index c1d51fb..11e181d 100644
--- a/src/engine/lexer.ts
+++ b/src/engine/lexer.ts
@@ -66,12 +66,18 @@ const SINGLE: Record<string, TokenType> = {
 function isDigit(ch: string): boolean {
   return ch >= '0' && ch <= '9';
 }
+/** Any Unicode letter, not just ASCII -- a table/sheet name (used unquoted
+ *  as a formula qualifier whenever it has no spaces) routinely contains
+ *  "ä"/"ö"/"ü"/"ß" or other accented letters in this German-first app;
+ *  rejecting those at the lexer level made every such name effectively
+ *  unusable as a bare qualifier (`Übersicht::A1` used to throw outright). */
+const IS_LETTER = /\p{L}/u;
 function isWordStart(ch: string): boolean {
   // '$' only ever appears inside a cell reference ("$A1", "A$1", "$A$1"), but
   // the lexer doesn't distinguish those from any other bare word (see the
   // module doc) -- it's just part of the word here, and the parser is the one
   // that validates/interprets it as a locked reference axis.
-  return (ch >= 'A' && ch <= 'Z') || (ch >= 'a' && ch <= 'z') || ch === '_' || ch === '$';
+  return IS_LETTER.test(ch) || ch === '_' || ch === '$';
 }
 function isWordPart(ch: string): boolean {
   return isWordStart(ch) || isDigit(ch);