Co-integration problem

[Bablowski]

Hello,

I would like to ask you for help with this problem. I have tested two series (non-stationary I(1) ) if they are co-integrated. Series of residuals from linear regression tested by ADF gives result of stationarity. This should mean that two tested series are co-integrated. Problem is that when I test if these two series are co-integrated by Engel-Granger single-equation test, the result is that they are not co-integrated because p-value is higher than 5% or 10%. So there is this discrepancy. Could you help me, please?