There is a double confusion here because "primary" and "secondary" were used to refer to IDE channels. Most machines had two IDE channels (i.e. physical connectors on the controller card), each of which could have a master and slave device (two connectors on the cable).
So you have IDE0, primary, and IDE1, secondary. For the four devices a typical system would support, they would be referred to as primary master, primary slave, secondary master and secondary slave. This was extremely accepted terminology.
Newer machines and BIOSes could usually boot from any of these four devices but originally, many machines could only boot the primary master. That's why it's the master and the other one is the slave -- it is subservient in the sense that it can not be a boot drive and was usually used for secondary storage, not OS.
So you have IDE0, primary, and IDE1, secondary. For the four devices a typical system would support, they would be referred to as primary master, primary slave, secondary master and secondary slave. This was extremely accepted terminology.
Newer machines and BIOSes could usually boot from any of these four devices but originally, many machines could only boot the primary master. That's why it's the master and the other one is the slave -- it is subservient in the sense that it can not be a boot drive and was usually used for secondary storage, not OS.