diff --git a/conda/recipes/mantiddocs/build.sh b/conda/recipes/mantiddocs/build.sh
index c9bd48a49212..f1de26c09ff2 100644
--- a/conda/recipes/mantiddocs/build.sh
+++ b/conda/recipes/mantiddocs/build.sh
@@ -37,7 +37,7 @@ cmake --build . --target StandardTestData
 export STANDARD_TEST_DATA_DIR=$SRC_DIR/build/ExternalData/Testing/Data
 echo 'datasearch.directories = '$STANDARD_TEST_DATA_DIR'/UnitTest/;'$STANDARD_TEST_DATA_DIR'/DocTest/' >> $PREFIX/bin/Mantid.properties
 
-# Set QT_APA_PLATFORM=offscreen so we do not need to run with xvfb. Xvfb hides a lot of debug output
+# Set QT_QPA_PLATFORM=offscreen so we do not need to run with xvfb. Xvfb hides a lot of debug output
 export QT_QPA_PLATFORM=offscreen
 cmake --build . --target docs-qthelp
 
diff --git a/docs/source/algorithms/CalculateDIFC-v1.rst b/docs/source/algorithms/CalculateDIFC-v1.rst
index 277956bc4863..d46ec692b7e2 100644
--- a/docs/source/algorithms/CalculateDIFC-v1.rst
+++ b/docs/source/algorithms/CalculateDIFC-v1.rst
@@ -18,7 +18,7 @@ an instrument.
 Or if OffsetMode is `Signed` :math:`DIFC` will be calculated with the following equation
 for logarithmically binned data:
 
-.. math:: DIFC = \frac{m_n}{h}&(L1 + L2)&2sin\theta & * & (1+|BinWidth|)^{-offset}
+.. math:: DIFC = \frac{m_n}{h} \cdot (L1 + L2) 2 \sin(\theta) \cdot (1+|BinWidth|)^{-offset}
 
 DIFC is used in the equation
 
diff --git a/docs/source/algorithms/ConvertDiffCal-v1.rst b/docs/source/algorithms/ConvertDiffCal-v1.rst
index c947aa6d6800..f9612bdb4e70 100644
--- a/docs/source/algorithms/ConvertDiffCal-v1.rst
+++ b/docs/source/algorithms/ConvertDiffCal-v1.rst
@@ -35,11 +35,11 @@ the following equations:
 
 Update existing calibration:
 
-.. math:: DIFC = DIFC_{old} & * & (1+|BinWidth|)^{-offset}
+.. math:: DIFC = DIFC_{old} \cdot (1+|BinWidth|)^{-offset}
 
 Calculate :math:`DIFC` from geometry of the experiment:
 
-.. math:: DIFC = \frac{m_n}{h}&(L1 + L2)&2sin\theta & * & (1+|BinWidth|)^{-offset}
+.. math:: DIFC = \frac{m_n}{h} \cdot (L1 + L2) 2 \sin(\theta) \cdot (1+|BinWidth|)^{-offset}
 
 The calculations for signed mode is appropriate for full-pattern cross-correlation with logarithmically binned data