Without knowing your exact experimental procedure it's difficult to say.
You should look at common sources of error, and see how they apply to your experiment.
Human error is a common example, and you've identified this when you said "reading off the value of the extension".
Other common sources of error:
Parallax - when taking a measurement where there is any distance between what's being measured and the scale being used, then if your eye is not in exactly the right place, the measurement you read can be distorted either way.
Resolution - All measuring instruments have a minimum unit to which they can measure to. E.g. a ruler might measure to the nearest 1 milimetre. This introduces an uncertainty in the measurement because your measurement could be anywhere between +/- 0.5 of the minimum unit. For example, say you measured the diameter of a ball to be 17mm, and your ruler only measured to the nearest mm, the actual diameter could be anywhere between 16.5 and 17.5mm, and you would still record it as 17mm, so it would have an uncertainty of +/-0.5mm. - This can occur with multiple measurements within the same experiment, not just with length measurements.
Other factors - think carefully about the assumptions you've made regarding your setup, and how these might not be perfect - these will be very specific to your setup, and the components you've used.