diff --git a/_posts/2024-04-01-cobol.md b/_posts/2024-04-01-cobol.md
new file mode 100644
index 0000000..023eb7f
--- /dev/null
+++ b/_posts/2024-04-01-cobol.md
@@ -0,0 +1,25 @@
+---
+title: COBOL
+code: |-
+       IDENTIFICATION DIVISION.
+       PROGRAM-ID. HFPMATH.
+       DATA DIVISION.
+       WORKING-STORAGE SECTION.
+       77 X-32 USAGE COMP-1.
+       77 X-64 USAGE COMP-2.
+       PROCEDURE DIVISION.
+           COMPUTE X-32 = 0.1 + 0.2.
+           DISPLAY X-32.
+           COMPUTE X-64 = 0.1 + 0.2.
+           DISPLAY X-64.
+           MOVE 0.1 TO X-64.
+           DISPLAY X-64.
+result: |-
+ 0.30000001
+ 0.3
+ 0.09999999999999999
+---
+
+COBOL encodes floating-point numbers using the [HFP][1] format instead of IEEE754, leading to different but analogous computation inaccuracies.
+
+[1]: https://en.wikipedia.org/wiki/IBM_hexadecimal_floating-point