Double and float are the same on Arduino (and indeed 4 bytes of data).
If it's just about the value, it's better to send it in binary format than in human readable format (saves conversions and much less data to transmit).
By doing a simple multiplication you can convert your float into int: *10 and you have the first decimal, good enough for most sensors which are far less accurate than that anyway. Then you can transmit your value in just two bytes.