Thanks but now this no longer makes sense: https://bitcoincore.org/en/segwit_wallet_dev/#transaction-serialization
"Each witness field starts with a compactSize integer to indicate the number of stack items for the corresponding txin"
Additionally there is fd0102 value indicating the size of the public key script in the SegWit transaction which is in fact a CompactInt not an OP_PushData which should have been 4d0102 indicating length of 513 bytes (which is available in the legacy transaction as an OP_PushData). Note that these transactions were created by Electrum and are already in (TestNet) blocks.
The witness data in a transaction (not to be confused with the scriptSig or scriptPubKeys) is not a script. It is a stack of bytes (can also be viewed as an array of an array of byte values). This is not a script, so it uses compact size uints, not opcodes."Each witness field starts with a compactSize integer to indicate the number of stack items for the corresponding txin"
Additionally there is fd0102 value indicating the size of the public key script in the SegWit transaction which is in fact a CompactInt not an OP_PushData which should have been 4d0102 indicating length of 513 bytes (which is available in the legacy transaction as an OP_PushData). Note that these transactions were created by Electrum and are already in (TestNet) blocks.
Compact size uints and pushdata opcodes happen to overlap for small values.
I've edited my above post as it was wrong. I misread your question.
