Double Precision Operands have a Tag 2 and occupy 2 Stack Cells.

001C (02,0006) 2 000000 000000 DOUBLE1 (Double) 2nd word:0 
001B (02,0005) 2 000000 000000 DOUBLE1 (Double) 1st word:0 , Double:0.0
 

The 1st Word is pushed onto the Stack, and then the 2nd Word is pushed.

They can be generated using any sequence of operators which result in a Tag 2 Word.

The Algol Compiler creates a DOUBLE Operand with this code sequence,

ZERO
XTND

Two Single Precision Operands can be combined to form a Double Precision Operand with the JOIN Operator.

LT8 <byte>
LT8 <byte>
JOIN

There are no special BlockExit requirements.

A Comparison between a Double Precision Operand and a Single Precision Operand using the SAME Operator always returns FALSE.