diff --git a/tests/test_base.py b/tests/test_base.py
index 080bf4c..aeab5d7 100644
--- a/tests/test_base.py
+++ b/tests/test_base.py
@@ -148,9 +148,10 @@ def test_vectorized_evaluate_scale(jit):
     cg = CorrectionWithGradient(schemas["scale"])
     evaluate = jax.jit(cg.evaluate) if jit else cg.evaluate
     x = np.array([0.0, 1.0])
-    values, grads = jax.value_and_grad(evaluate)(x)
+    values, grads = np.vectorize(jax.value_and_grad(evaluate))(x)
+    assert len(values) == len(x)
     assert np.allclose(values, [1.234, 1.234])
-    assert len(grads) == 2
+    assert len(grads) == len(x)
     assert grads[0] == 0.0
     assert grads[1] == 0.0
 
@@ -162,10 +163,12 @@ def test_vectorized_evaluate_simple_uniform_binning():
     values = cg.evaluate(x)
     # here and below, the magic numbers have been checked by plotting
     # the bins and their contents, the corresponding spline, and its derivative.
+    assert len(values) == 3
     assert np.allclose(values, [3.47303922, 5.15686275, 1.0])
 
     grads = np.vectorize(jax.grad(cg.evaluate))(x)
     expected_grad = [0.995098039, 0.0, 0.0]
+    assert len(grads) == len(expected_grad)
     assert np.allclose(grads, expected_grad)
 
 
@@ -174,8 +177,10 @@ def test_vectorized_evaluate_simple_nonuniform_binning():
     x = [3.0, 5.0, 11.0]  # 11. overflows: it tests clamping
 
     values = cg.evaluate(x)
+    assert len(values) == 3
     assert np.allclose(values, [2.0, 3.08611111, 1])
 
     grads = np.vectorize(jax.grad(cg.evaluate))(x)
     expected_grad = [0.794444444, 0.0, 0.0]
+    assert len(grads) == len(expected_grad)
     assert np.allclose(grads, expected_grad)