Time bases (such as Crystal oscillators in lab equipment) suffer from a range of error sources.

(1) Drift causes the oscillator frequency to shift over time, mainly caused by temperature changes

To avoid drift, the main thing that needs to be done is to counteract the effect of temperature. This can be done passively by allowing the time base to reach operating temperature prior to usage, by using an external time base that is built to internally offset temperature differences (TCXO), by using a heated time base (OCXO), or by switching to an external high precision source directly (such as a radio station or a rubidium oscillator). 

(2) Accuracy causes the Time base to produce a timing signal which has a higher or lower frequency than the real reference value

This can only be avoided by directly or indirectly linking to a signal which is defined to be accurate, in other words an atomic clock. Surprisingly this is simpler than one would imagine!

(3) Synchronisation errors occur if different lab equipment is used together which uses individual (internal) time bases. Accuracy in such a setup will be limited to the weakest link in the chain.

The first step here should be to link the equipment time bases together if possible or at least regularly manually calibrate the instruments to a master source.

Multiple CW transmitting radio stations exist that have a transmit frequency tied to an atomic clock. In Europe, the closest station in the shortwave range would be Russian station RWM at 4996 KHz. The USA has WWV. Find a station for your region in this list.

You can use this to validate the correctness of a time base in a radio by tuning the station in CW mode and checking that the radio sees it at 4996 KHz. Another method is to set the dial to SSB and then tuning the radio until the signal becomes deeper and deeper in pitch, then finally just disappears, then checking the dial.

All these methods have one common disadvantage which is that the tuning relies on hearing and is as such limited. With a lot of luck, one might be able to achieve 1 ppm of accuracy (5 Hz on the RWM signal). Still, for most receivers, this is good enough as their dials do not have a higher resolution anyway.

A better method to derive a reference signal from the radio signal is by repeatedly filtering and buffering it without actually downconverting (as the LO would add to the inaccuracy). I do this with the German time standard station DCF77 which generates an atomic clock locked signal at 77.5 KHz which can be received in many places in Europe. The receiver first generates a 77.5 MHz TTL signal, then uses a PLL to tune a VCO to exactly 10MHz, based on the received signal.

A similar device can be purchased from the shop of German magazine "Funkamateur". Schematics are published there. 

The one big advantage that this device has is that it works almost everywhere, even in the basement, as long as you remain close enough to the transmitter.

Each one of the GPS satellites transmits a precision clock signal which can be received with fairly little effort. The principle of operation is very similar to the previous. A 10KHz signal locked to the satellites clock is produced by a ready made GPS board, then the same is compared to a divided sample of a 10MHz clock.

The work here has been done by Brooks Shera, who published an excellent article on the subject in QST magazine. The article, Schematics and software can be found on Brooks website.

Similar concept and very excellent build quality is the work of James Miller G3RUH