1. In the degenerate case C(C0 in the new version), it may occur that $t_1=0$ or $t_2=0$.
Such cases should have been considered separately. It turns out such cases are similar to Case B.

2. In degenerate cases in Section 4, we apply Lemma 3.3, which is essentially a special case of Theorem 2 of Laurent, Acta Arith. 133 (2008), 325--348.
In the previous version, we calculated constants by choosing $\log\alpha_i$ in Lemma 3.3 such that $\abs{\Im \log\alpha_i}<\pi/2$.
However, this is not allowed in degenerate cases.
For example, in Case B.i, $\log\alpha_4$ should be $b_2^\prime\log\alpha_2-b_1^\prime\log\alpha_1$ in order that
$d\log\alpha_4-b_3\pi i/2=\Lambda$ holds. If we take $\log\alpha_4$ to be its principal value, then
the new linear form $d\log\alpha_4-b_3\pi i/2$ may take different value from $\Lambda$ by a multiple of $\pi i$.
Thus we reestimated the absolute values $\abs{\log\alpha_i}$.


Other errors:

3. In Case B, it is $\varphi(1)$ and $\varphi(2)$, not $\varphi(1)$ and $\varphi(3)$ which are exchangable.

4. In Case B.b, we choose $\log \alpha_4^\prime$ such that $\alpha_4^\prime\in \{\pm \alpha_4, \pm \alpha_4 i\}$
and $0\leq \Im \log\alpha_4^\prime<\pi/2$ and rewrite $\Lambda$ as
$\Lm=d(b_2^\prime \log\alpha_2-b_1^\prime\log\alpha_1)-b_3\pi i/2=d\log \alpha_4^\prime-b_3^\prime \pi i/2$.
Then we can see that $\abs{b_3^\prime}\leq b_3$ and can continue the argument as in the previous version.

5. We removed impossible cases in Case B. For example, in Case B.i, 

6. Case I of Section 5: We made $KL$ explicit.

7. Case II of Section 5: We added some coefficients which were absent in (85)(86) (-> (118)(119)).

8. We added the reference to Lemma 3.4.