The measurement was made in my office, 3m from the lan cable tray which runs along the corridor outside, using a 0.9m active loop. This has a near flat response (+/-1dB) from 9kHz to 30MHz. The antenna factor curve was loaded into the measurement receiver. The field strength is represented as the equivalent E field, using 51.5dB as the conversion factor from H field. The discrete peaks are not the lan, they are intentional broadcast carriers.
The lan uses Hewlett Packard type Procure switch 2626 hardware, although I think older hardware is also used in the network. As you say, it's not possible to switch the system off for a background noise check, but I have determined that the local 160 and 80m interference at least is definitely generated by the lan system. The level stays pretty constant, i.e. the cumulative level does not vary day/night or weekday/weekend, so seems independent of data throughput.
This is another plot I did with the loop antenna attached to the
cable tray, i.e. zero measurement distance, so it shows the lan radiated spectrum more clearly. The noise floor of the measurement system is limited by the loop/preamp and is equivalent to 30dBuV/m,